Eighty-five Coffea canephora genotypes were evaluated for three crop seasons under irrigation in the Cerrado of the Distrito Federal, Brazil, to characterize the population regarding phenotype, yield and maturation cycle. The experiment was conducted through the Basic Model of Repeatability without Design methodology. Regarding the cycles, genotypes were divided into four groups with the following periods, in days, from resumption of irrigation to the cherry stage: very early (243-255), early (256-267), medium (268-280), and semilate (281-293). For yield, the repeatability obtained was approximately 33%, a medium value that represents the proportion of the permanent phenotypic variance in relation to the total phenotypic variance. The average repeatability obtained surpassed 59%, and accuracy, with three crop seasons, reached 77%. The selection gains obtained may be higher than 38%. There are promising genotypes for cultivation in the region under study. The mean repeatability and the accuracy obtained favor selection based on phenotypes.