The introduction of a new cultivar to a crop simulation models requires the estimation of cultivar coefficients that define the growth and development characteristics. The objective of this study was to employ pattern recognition approach to estimate cultivar coefficients to be used with CSM-CERES-maize model from least available measured data. The main goal of pattern recognition is to classify groups of data based on a priori knowledge extracted from the patterns. The approach that was used in this study is based on maize crop growth and development features including anthesis and harvest maturity dates, maximum leaf area index, above ground biomass and grain yield. Based on the similarity measure as central calculation of pattern recognition the algorithm searched the space of features of other cultivars in the database to find the most similar as the best match to the target cultivar. To construct the feature database, 6935 hypothetical cultivars by combining different values of six maize cultivar coefficients were constructed. CSM-CERES-Maize model was run at potential production for all these cultivars and the outputs were used as feature database. Experimental data from Gainesville, Iowa, and Spain were used in this study. To verify our model we used 28 different maize cultivars available in the pool of DSSAT database. The model was run for all 28 cultivars under the three study sites conditions and both the simulated coefficients and crop data, simulated based on these coefficients, were compared with original crop growth data (DSSAT database). Both correlation coefficient (r) and Root Mean Square Difference (RMSD) confirmed that our approach provided reliable estimates of maize cultivar coefficients. RMSD values were different across different sites but still indicated the level of accuracy was targeting for those own no more than one year experimental data and demand the best possible initial guess