Climatic variability and extremes climatic events are responsible for one-third of the global variability in crop yields. Water shortage, caused by large variations in the amount, frequency and timing of rainfall during the crop cycle, is one of the major abiotic stresses limiting crop production. In this context, environmental variability creates significant challenges for farmers as optimal management practices vary from season to season. The aim of this study was to optimize wheat (Triticum aestivum L.) management practices considering rainfall variability in the west Pampas of Argentina. This region is characterized by predominantly summer rainfall with probable deficits in winter and early spring, and with high rainfall annual variability. Two genotypes, commonly used by farmers, with different time to flowering (LF: late flowering; EF: early flowering), under three initial soil water conditions (well-watered, moderatelywatered and dry conditions) at three sowing dates were simulated by DSSAT (CERES-WHEAT) for a series of 39 years. As expected, wheat yield was, in average, higher for the LF than the EF genotype (5437 vs. 5112 kg ha-1 for LF and EF, respectively; p<0.05). Yield differences were even more evident under no-water-restrictions due to a higher resource capture (7915 vs. 6956 kg ha-1 for LF and EF, respectively; p<0.05). However, in dry years (less than 200 mm) opposite results were observed with the EF genotype yielding more than the LF genotype (4128 vs. 3255 kg ha-1 for EF and LF, respectively; p<0.05). Variations in grain yield were associated with changes in grain number (r2 = 0.80) rather than grain weight, which remained stable across different environmental conditions. Reductions in grain number were a consequence of a lower partition to reproductive organs (p<0.05), principally limited by water shortage. The LF genotype, with a longer vegetative phase, used more water than the EF genotype before anthesis and therefore depleted soil water reserves (p<0.05) that were scarce for reproductive phases. Differences between genotypes were more evident under initial soil well-watered conditions (16%) rather than moderately-watered (13%) or dry conditions (9%). Sowing date did not show any effect nor interaction with the other treatments. Therefore, by better matching crop cycle length with changed rainfall distribution, genotype selection can partially compensate the deleterious impacts of water deficit on rainfed wheat yield in the West Pampas of Argentina. Results highlight the need to design management practices in a scenario-dependent manner in order to reduce yield gaps.