Given the importance that yellow grain maize hybrids have for human and animal consumption, the Maize Program of the Instituto de Ciencias y Tecnologia Agricola (ICTA) initiated a project using 10 maize lines to genera te double (DC) and three way cross (TWC) hybrids which would be superior to the commercial hybrid HA-46. Most of the 10 lines had different genetic origins. Specific objectives were: a) estimate specific and general combining ability (SCA, GCA) for each of the simple crosses; and b) to predict yield for the DC and TWC hybrids. A 7x7 simple lattice design was used to evaluate 42 simple crosses and 7 checks. SCA and GCA were estimated using the Fourth Griffing's Design, and yield predictions using Jenkins Method B. Two trials were established in Cuyuta and La Maquina (40 to 150 meters above sea level). Results obtained indicated that the two null hypothesis could not discarded: 1) the variance for GCA (g 2) was considered to be equal to zcro givcn that the testers wcre previously selected; and the variance for SCA (s 2) was different from zero, possibly because some testcrs had different origins while others were derived from the same source. 2) SCA was as important as CGA in predicting the yicld of the simple crosses. Predicted yield levels for DC’s and TWC's were highcr than the predicted yield for HA-46. These results suggest that when these DC's and TWC's are formed they will be superior to HA-46 and as such fulfill the general objective set forth.