The critical value of soil Olsen-P is the point above which the probability of crop yield response to fertilizer P is small or nil. Determining this critical value is fundamental when making appropriate P fertilizer recommendations. In this study, the critical values were determined for continuous maize (Zea mays L.)-winter wheat (Triticum aestivum L.) cropping systems from a 15-year field experiment across three sites in China using linear-linear, linear-plateau and Mitscherlich models. The mean critical values for maize using the three models ranged from 12.1 to 17.3 mg P kg −1 (average 15.3 mg P kg −1 ) and for winter wheat from 12.5 to 19.0 mg P kg −1 (average 16.3 mg P kg −1 ) among study sites. The mean critical value for maize was approximately 7% lower than that for winter wheat across all sites based on the three models. Critical values identified by the Mitscherlich model were 1.4 to 2.1 times those from linear-linear and 1.3 to 1.9 times of those from linear-plateau and were crop and site dependent. There was a significant negative correlation (P<0.05) between the mean critical value from the three models and the observed P uptake by either maize or wheat. Our study shows that the critical values can vary with sites, crops and models used, and thus caution should be taken when selecting the most appropriate one when making P fertilizer recommendations for agronomic return and to minimize chances of negative environment impact from overfertilization.