Males of the coreid bug,Margus obscurator(F.), were individually numbered in the field in southeastern Georgia (USA) and censused daily for the 6 weeks in spring from the initiation of mating activity until the complete decline of the adult population. Data was analyzed for the subset of males sighted at least 5 times since the relationship between number of observations and variance in mating success leveled off at five observations. Variance in male mating success exceeded that generated by null models, suggesting a phenotypic determinant of mating success. Body length was significantly positively correlated with male mating success. Longer males were more likely to retain a grasp on females which resist, by fleeing, most courting males. The intensity of sexual selection (standardized selection differential) on male size was greater in the second half of the season (0.43 versus 0.21) when both host plant abundance and the proportion of females in the population had declined. Consequently, male density and the number of intermale aggressive interactions increased. During fights, larger males were more likely to retain access to females or the flower heads on which mating occurred. The opportunity for sexual selection (= squared coefficient of variation for mating success) was also greater in the second half of the season (0.90 versus 0.20). The opportunity for sexual selection increased 2.5 times faster than the intensity of sexual selection on size between halves of the season, reflecting the greater male-bias in the operational sex ratio (proportion of males = 0.57 versus 0.49), a more patchy distribution of females and, perhaps, reflecting the operation of additional components of phenotypic selection.