Field experiments were conducted in the irrigated and the rain-fed crops of 2000 and 2001 at Suwan Farm, Pak Chong, Nakon Ratchasima, Northeastern Thailand to determine the spatial dispersion of cotton bollworm, Helicoverpa armigera (Hubner) larvae on four cotton varieties/lines, namely, AP1 and AP2, the mutant lines, SR60 (Sri Samrong 60) and SD1 (Sarid 1), the recommended varieties. RCB was used with four replicates for both crops. Spatial dispersion was analyzed using the variance-to-mean ratio (s **(2)/x mean), Morisita Index (I delta) and the negative binomial parameter (k). Although cotton bollworm larvae exhibited mainly clumped distribution, sometimes it was found to be randomly disperse during the sampling periods of both crops. In general, H. armigera larvae were at high aggregation (k=0.10) on the irrigated crop and low aggregation (k=3.92) on the rain-fed crop. Dispersion information was used to select an optimum sample size at the 0.05 probability. Optimum sample size of 15 and 30 plants per 140 square m were needed to monitor low and high aggregation levels of H. armigera larvae, respectively.