The frequency of phenotypic resistance to herbicides in previously untreated weed populations and the herbicide dose applied to these populations are key determinants of the dynamics of selection for resistance. In total, 31 Lolium rigidum populations were collected from sites with no previous history of exposure to herbicides and where there was little probability of gene flow from adjacent resistant populations. The mean survival frequency across all 31 populations following two applications of commercial rates ( 375 g ha(-1)) of the acetyl- coenzyme A carboxylase ( ACCase) inhibiting herbicide, diclofop- methyl was 0.43%. Survivors from five of these populations were grown to maturity and seed was collected. Dose - response experiments compared population level resistance to diclofop- methyl in these selected lines with their original parent populations. A single cycle of herbicide selection significantly increased resistance in all populations ( LD50 R: S ratios ranged from 2.8 to 23.2), confirming the inheritance and genetic basis of phenotypic resistance. In vitro assays of ACCase inhibition by diclofop acid indicated that resistance was due to a non- target- site mechanism. Following selection with diclofop- methyl, the five L. rigidum populations exhibited diverse patterns of cross-resistance to ACCase and ALS- inhibiting herbicides, suggesting that different genes or gene combinations were responsible for resistance. The relevance of these results to the management of herbicide resistance are discussed.