- Recurrent selection programmes use one or just a few selection criteria, however other indirect traits may be unpredictably changed in a population. This study was conducted to determine the indirect effects of 10 cycles of bi-directional recurrent selection for cotyledon size in the model system of rapid-cycling Brassica rapa. Eight sub-populations (four large- and four small-cotyledon sizes) were phenotypically and cytologically evaluated. Each sub-population was measured by five phenotypic traits. Ploidy and pollen viability were studied and compared with the initial population. Total weight was significantly different in broad and bottleneck-1 sub-populations. Total cell number exhibited statistically significant differences in broad and bottleneck-1 sub-populations, while cell number per unit area exhibited statistically significant differences in broad, bottleneck-2 and bottleneck-3 sub-populations. Decreases in pollen viability in comparison with the base population were observed in three sub-populations. Among the eight sub-populations studied, the most significant phenotypic differences were observed within broad sub-populations. Based on the above, it is possible that bi-directional recurrent selection for cotyledon size may have been a result of indirect selection for two processes, endoreduplication and cell division.