The objective of this study was to determine the main sources of variation in herbage densities of infective third stage larvae of goat strongyles during the marked dry season of 1994 in Guadeloupe (FWI). Herbage samples were collected for L3 density (LD) determination by an accurate method, 4 times at 4-week intervals in 58 paddocks of 21 farms spread out in five regions of the archipel of Guadeloupe. At the same time, EEC of each grazing animal and fecal culture for parasite genus determination according to sex and age were carried out. Stocking rate, dry matter content of soil, and daily climatic data were also recorded. An index of egg development in larvae (IEDL) was calculated as the ratio of LD to the eggs deposed during the 4th, 3rd, and 2nd weeks before sampling. Medians of LD in herbage were 3397, 1853, 1410, and 324 L3/kg DM for all parasites, Haemonchus, Trichostrongylus, and Oesophagostomum, respectively. Date of sampling, region, and irrigation practice in the northern windward region were the main sources of variation in LD and in frequency of each parasite. LD decreased as the dryness lasted, but it remained important (500 L3/kg DM) despite the drought. LD in windward regions were higher than in other regions. The region, the farm, and the paddock were the main sources of variation of IEDL. LD of each parasites were inversely correlated to global radiation recorded I to 3 weeks before herbage sampling, but no relation was found with rainfall data . Trichostrongylus frequency in L3 population increased as the dryness lasted. A dryness axis was extracted from environmental variables (climatic data, dry matter of soil, duration of dryness) by a multiple factorial procedure. LD and Haemonchus frequency in L3 population were inversely correlated to dryness axis (p < 0.01). In contrast, Trichostrongylus frequency was positively correlated to the dryness component.