The uncertainties in sea ice extent (total area covered by sea ice with concentration >15%) derived from passive microwave sensors are assessed in two ways. Absolute uncertainty (accuracy) is evaluated based on the comparison of the extent between several products. There are clear biases between the extent from the different products that are of the order of 500 000 to 1 x 10(6) km(2) depending on the season and hemisphere. These biases are due to differences in the algorithm sensitivity to ice edge conditions and the spatial resolution of different sensors. Relative uncertainty is assessed by examining extents from the National Snow and Ice Data Center Sea Ice Index product. The largest source of uncertainty, similar to 100 000 km(2), is between near-real-time and final products due to different input source data and different processing and quality control. For consistent processing, the uncertainty is assessed using different input source data and by varying concentration algorithm parameters. This yields a relative uncertainty of 30 000-70 000 km(2). The Arctic minimum extent uncertainty is similar to 40 000 km(2). Uncertainties in comparing with earlier parts of the record may be higher due to sensor transitions. For the first time, this study provides a quantitative estimate of sea ice extent uncertainty.