For a number of generations, we followed the frequency, performance and fecundity of an initial mixture (50[ratio ]50) of triazine resistant (R) and susceptible (S) but otherwise near-isogenic lines of Brassica rapa (for 3 yr) and of Chenopodium album (for 2 yr) in neighbouring experimental garden plots, which differed in aspect (north-facing versus south-facing), shading, and transparent cover from precipitation. Each of eight treatments was replicated four times per species. Seed from each plot was kept separate and sown back in its plot of origin to provide the next generation. For both Brassica and Chenopodium, R frequency changed between generations. In Brassica R frequency declined consistently from an initial 0.5 and, after 2 yr, R plants had disappeared in all replicates in two treatments and averaged <0.126 in the remaining treatments. However, R frequencies in the smaller second generation samples in each year were less consistent. In Chenopodium, R frequency increased consistently from an initial 0.5 in all treatments and replicates during 1995 to average 0.772 on south-facing sites and 0.675 on north-facing sites. However, in 1996 Chenopodium R frequency did not change significantly from that in 1995 for any treatment. Brassica S plants were more fecund that R plants, except in shaded conditions, and south-facing plots produced significantly more S seeds than north-facing plots. Both Brassica plants and plots were more productive in 1996 than in 1995. By contrast, Chenopodium was much less productive in 1996, a season with more climatic extremes between plots than 1995. In 1995, R plants were more fecund than S plants in south-facing plots and in covered conditions, but this difference was not detected in 1996. South-facing Chenopodium plots were more productive than north-facing plots in 1995, but the reverse was the case in 1996. Cover resulted in an increase in seed production in Chenopodium on south-facing plots in 1995, but a decline in 1996. We conclude that selection might act against triazine-resistant plants in B. rapa under the climatic conditions in the UK, but in C. album, which is more susceptible to extreme conditions of light and water stress, triazine-resistant phenotypes might predominate in low stress cool moist conditions. We suggest that a relationship might exist between the overall climatic tolerance of a species, and climate thresholds beyond which herbicide-resistant phenotypes might spread. Thus, in the absence of herbicide, it is not inevitable that the frequency of R phenotypes will decline in field conditions.