Genetic and agronomic contributions to yield gains for wheat (Triticum aestivum L.) in the Yaqui Valley of northwest Mexico were estimated for the period from 1968-1990. Five problems associated with estimating sources of yield gains were considered, namely: (1) adjusting yield gains for variation in weather over the study period; (2) considering annual cultivar-by-weather interactions, which are potentially problematic when relative yields of cultivars are generated over only part of the study period; (3) overestimating relative yields of cultivars if the check cultivar(s) become susceptible to disease; (4) ensuring that yields as estimated from research station trials represent cultivar performance under farmers' conditions; and (5) allowing for cultivar-by-management interactions. With these factors considered, 28% of the weather-adjusted yield gain of 103 kg ha - 1 y - 1 was attributed to genetic gain (i.e., cultivar improvement). If the yield gains had not been adjusted for annual weather variation, genetic gain would have been overestimated at 50%. By contrast, 48% of the gain was attributed to increased use of N fertilizer, driven at least in part by a decline in N prices; no N-by-cultivar interactions were apparent. The remainder of the yield gap (24%) could not be attributed to specific factors, although P application rates increased over time, and negative deviations from linearity were associated with years having greater than normal rainfall in November and December (a phenomenon which can delay sowing and/or reduce crop stand). In addition to identifying factors contributing to yield gains, technologies that have contributed benefits not measured by yield are discussed.