I spent two weeks trying to pull pricing tables out of 40+ vendor PDFs for a comparison spreadsheet. Copy-paste gave me mangled columns. Online converters choked on merged cells. Here’s what actually worked after testing 9 different tools.
If you just need a free PDF editor for basic tasks, that’s a different problem. Extracting tables specifically – where columns, rows, and cell boundaries matter – requires tools built for structured data extraction.
Quick Comparison: Best PDF Table Extraction Tools
| Tool | Type | Best For | Price | Output Formats | Handles Merged Cells |
|---|---|---|---|---|---|
| Tabula | Desktop (Java) | Clean, text-based PDFs | Free | CSV, TSV, JSON | Partial |
| Smallpdf | Online | Quick one-off extractions | Free (2/day), $12/mo | Excel, CSV | Yes |
| iLovePDF | Online | Batch processing | Free (limited), $4/mo | Excel | Yes |
| Camelot | Python library | Automation, developers | Free | CSV, Excel, HTML, JSON | Yes (lattice mode) |
| Google Docs | Online | Quick and dirty | Free | Copy-paste | No |
| Adobe Acrobat Online | Online | Scanned PDFs (OCR) | Free (limited), $12.99/mo | Excel, Word | Yes |
| PDF2Go | Online | No signup needed | Free (limited), $6.50/mo | Excel, CSV | Partial |
| Excalibur | Web UI (local) | Visual table selection | Free | CSV, Excel, JSON | Yes |
| Microsoft Excel | Desktop | Windows users who already have it | Included with Microsoft 365 | Native Excel | Partial |
1. Tabula – Best Free Desktop Option
Tabula is the tool I keep coming back to. It’s open-source, runs locally (your files never leave your machine), and handles most text-based PDFs without issues.
How it works: You upload a PDF, draw a box around the table you want to extract, and Tabula pulls the data into rows and columns. It has two detection modes – “Stream” for tables without clear borders and “Lattice” for tables with visible gridlines.
I extracted 40 pricing tables using Tabula’s Lattice mode and got clean results on 34 of them. The 6 failures were all scanned PDFs (Tabula can’t do OCR) or tables with irregular merged cells spanning 4+ columns.
What I like:
- Completely free, no limits on file count or size
- Data stays on your computer
- Batch processing – select multiple tables at once
- Export to CSV, TSV, or JSON
Where it falls short:
- Requires Java installed on your system
- Can’t handle scanned/image-based PDFs at all
- Struggles with complex merged cells
- UI feels dated (it works, but it’s not pretty)
Platform: Windows, macOS, Linux. Requires Java 7+.
2. Smallpdf – Best Online Tool for Quick Extractions
When I need to grab a table fast and don’t want to install anything, Smallpdf is my first stop. Upload the PDF, and it automatically detects tables and converts them to Excel format.
The free tier gives you 2 document conversions per day. For most people doing occasional table extraction, that’s enough. The accuracy surprised me – on a test with a 15-column financial report, Smallpdf preserved the column alignment better than Tabula did.
What I like:
- No installation, works in any browser
- Good automatic table detection
- Handles some merged cells correctly
- Clean Excel output with formatting preserved
Where it falls short:
- 2 free operations per day is tight if you have many files
- Files are uploaded to their servers (privacy concern for sensitive documents)
- Occasional errors with tables that span multiple pages
Price: Free (2/day), Pro at $12/month.
3. iLovePDF – Good for Batch Processing
iLovePDF offers a PDF editing suite with table extraction built in. The free tier handles smaller files, and the paid plan at $4/month is cheaper than most competitors.
The table extraction accuracy is comparable to Smallpdf. Where iLovePDF wins is batch mode – you can upload multiple PDFs and process them all at once, which saved me about an hour when I needed to extract data from 20 quarterly reports.
One thing that bugged me: the free version adds a watermark to the output file header. It doesn’t affect the data, but it looks unprofessional if you’re sharing the spreadsheet directly.
What I like:
- Batch upload and processing
- Desktop app available for offline use
- Cheapest paid option at $4/month
Where it falls short:
- Free version has file size limits (15MB)
- Watermark on free conversions
- Less accurate than Smallpdf on complex layouts
4. Camelot – Best for Developers and Automation
If you’re comfortable with Python, Camelot is probably the most powerful free option available. It’s a library that extracts tables from PDFs and gives you granular control over how the extraction works.
Installing it takes a few steps (you need Ghostscript and Tkinter as dependencies), but once it’s set up, you can process hundreds of PDFs with a simple script. I wrote a 12-line Python script that extracted tables from 200+ contract PDFs and exported them as individual CSV files in under 10 minutes.
Camelot’s Lattice mode is exceptionally good at recognizing table boundaries. It uses the actual lines drawn in the PDF to identify cells, which means it handles merged cells better than most tools. Stream mode works on tables without borders but is less reliable.
Install and basic usage:
pip install camelot-py[cv]
import camelot
tables = camelot.read_pdf('report.pdf', pages='all')
tables[0].to_csv('output.csv')What I like:
- Full automation capability
- Lattice mode accuracy is the best I’ve tested
- Output to CSV, Excel, HTML, JSON, or SQLite
- Accuracy reports tell you how confident the extraction was
Where it falls short:
- Requires Python knowledge
- Dependency installation can be frustrating on Windows
- No OCR – text-based PDFs only
5. Google Docs – Quickest No-Install Method
Here’s the thing about Google Docs: it’s terrible at table extraction. But it’s free, everyone has it, and sometimes “terrible” is good enough.
Upload the PDF to Google Drive, right-click, open with Google Docs. It’ll try to convert the content, and simple tables often come through with the structure mostly intact. You can then copy the table into Google Sheets.
I tested this on 10 PDFs. Results: 4 clean extractions, 3 partial (needed manual fixing), and the remaining ones came out as scrambled text. If your table is simple – no merged cells, no spanning headers, consistent column widths – Google Docs works fine. For anything more complex, use a dedicated tool.
Best for: Simple tables in text-based PDFs when you don’t want to sign up for anything new.
6. Adobe Acrobat Online – Best for Scanned PDFs
Adobe’s free online tool at acrobat.adobe.com lets you convert PDFs to Excel. The main advantage over every other tool on this list: it has built-in OCR. If your PDF is a scan of a printed table, Adobe is the only free option that can reliably extract the data.
I tested it with a scanned invoice that had a 7-column table. Adobe’s OCR recognized every cell correctly, including currency symbols and decimal values. Tabula couldn’t even see the table. Camelot returned empty results.
The free tier limits you to a handful of operations before asking you to sign in, and then caps at about 2 conversions. Adobe Acrobat Pro costs $12.99/month, but if you already have a Creative Cloud subscription, you’ve got it included.
What I like:
- OCR handles scanned documents
- Extremely accurate formatting preservation
- Direct export to Excel or Word
Where it falls short:
- Very limited free tier
- Requires Adobe account
- Expensive if you need it frequently
7. PDF2Go – No Account Needed
PDF2Go converts PDFs to Excel without requiring you to create an account. Upload, convert, download. That simplicity is its main selling point.
Extraction quality is middle-of-the-road. It handled standard tables fine but struggled with the same complex layouts that trip up most online tools. The advantage is speed – the conversion typically finishes in under 30 seconds even for longer documents.
Price: Free with limits, Premium at $6.50/month.
8. Excalibur – Visual Selection with Camelot’s Engine
Excalibur is basically a web interface for Camelot. If you want Camelot’s accuracy but don’t want to write Python code, Excalibur gives you a visual tool where you draw boxes around tables, tweak detection settings, and export results.
It runs locally on your machine through a browser interface. Setup requires Python and a few commands, but once running, it’s straightforward. I found the visual table selection especially useful for PDFs with multiple tables per page where automatic detection picks up headers or footers as table data.
Best for: Non-developers who want Camelot-level accuracy with a point-and-click interface.
9. Microsoft Excel – Built-in Import (Microsoft 365)
If you have Microsoft 365, Excel can import data directly from PDFs. Go to Data > Get Data > From File > From PDF. Excel detects tables automatically and lets you preview them before importing.
The accuracy varies. Clean, well-structured tables import perfectly. Complex layouts with nested headers or merged cells often need manual cleanup. But since you’re already in Excel, fixing column alignment takes a few clicks rather than re-importing through another tool.
This feature isn’t available in older Excel versions (pre-2019) or the free online version. You need the desktop app with a Microsoft 365 subscription.
Tips for Better Table Extraction Results
After extracting tables from over 200 PDFs across these tools, here’s what I’ve learned:
Check if your PDF is text-based or scanned. Open the PDF and try to select text with your cursor. If you can highlight individual words, it’s text-based and most tools will work. If you can’t select anything, it’s a scanned image and you need OCR (Adobe Acrobat or a dedicated OCR tool first).
Pre-process messy PDFs. If a table spans multiple pages, try extracting specific pages first, then run the extraction on individual pages. Multi-page table extraction is where most tools fail.
Use Lattice mode when available. If the table has visible gridlines, always choose Lattice/bordered mode over Stream/automatic. The accuracy difference is substantial – I saw 85%+ accuracy with Lattice vs 60% with Stream on the same document.
Verify the output. No tool gets it right 100% of the time. Always spot-check at least the first and last row, plus any rows with unusual formatting (currency, dates, merged cells).
Try multiple tools. If one tool mangles a specific table, another might handle it perfectly. I’ve had cases where Tabula failed completely on a table that Smallpdf extracted flawlessly, and vice versa.
Which Tool Should You Pick?
For most people doing occasional table extraction: start with Smallpdf. It’s fast, free for light use, and accurate enough for standard tables.
If you have sensitive documents: use Tabula. Everything stays on your machine.
If you’re dealing with scanned PDFs: Adobe Acrobat Online is the only free option with decent OCR.
If you need to process dozens or hundreds of files: learn Camelot. The time investment pays off quickly.
For the full range of PDF operations – editing, merging, splitting, and converting – check our roundup of the best free PDF editors.
FAQ
Can I extract tables from scanned PDFs for free?
Adobe Acrobat Online offers limited free OCR-based table extraction from scanned PDFs. For unlimited free processing of scanned documents, you’d need to run OCR separately (using a tool like Tesseract) to convert the scan to a text-based PDF, then extract tables with Tabula or Camelot.
What’s the most accurate free PDF table extraction tool?
For text-based PDFs with visible gridlines, Camelot’s Lattice mode consistently gave me the best results in testing – around 90% accuracy on well-formatted tables. For online tools, Smallpdf performed best overall. Accuracy depends heavily on how the PDF was created, so testing 2 tools on your specific document is always worth the extra minute.
Can I extract tables from password-protected PDFs?
You need to remove the password protection first. Most tools on this list won’t process locked PDFs. You can unlock the PDF for free first, then run the table extraction.
How do I extract a table that spans multiple PDF pages?
Most tools extract tables page by page, so a table split across pages produces two separate outputs. Your best option is to extract each page’s portion separately, then merge the results in Excel or Google Sheets. Camelot with the pages=’all’ parameter handles some multi-page tables, but results vary depending on whether headers repeat on each page.
Is there a way to automate PDF table extraction?
Yes. Camelot (Python) and Tabula (Java, with tabula-java) both support scripting. You can write a script that processes a folder of PDFs and exports all tables to individual CSV or Excel files. For Camelot, the basic automation script is about 10 lines of Python code.