Wheat is a staple crop throughout much of India, but in many areas it is commonly sown past the optimum window for yields. Recent technologies, such as adoption of no-till practices or earlier maturing cotton and rice varieties, have enabled some farmers to sow wheat earlier, but repeatable and publicly available measurements of sow date trends are lacking. Here we utilize satellite measurements since 2000 to estimate sow dates over a decade throughout wheat growing areas in India. Comparisons with ground-based sow dates in Punjab confirmed the reliability of satellite estimates, and data from two independent satellite sensors were used as a robustness check. We find statistically significant (p < 0.05) shifts toward earlier sowing of wheat throughout much of Haryana and Uttar Pradesh, with insignificant changes in Punjab. A production-weighted average of the entire region indicates that, on average, wheat was sown 1 week earlier by 2010 than it was at the beginning of the decade. Using previously published experimental estimates of yield gains from earlier sowing, we estimate that an overall yield gain of at least 5% averaged across India can be explained by the sow date trend. Given that national yield changes since 2000 have been less than 5%, our results indicate that the sow date shift has been a major factor in yield changes over the past decade, and that the net yield effect of all factors other than sow date has been close to zero, perhaps even negative. The results also indicate that sow dates in much of Haryana and western Uttar Pradesh are nearing or already at the optimum window for yields, so that yield benefits from sow date shifts will likely diminish in the next decade.