BackgroundMosquito larval source management (LSM) is likely to be more effective when adequate information such as dominant species, seasonal abundance, type of productive habitat, and land use type are available for targeted sites. LSM has been an effective strategy for reducing malaria morbidity in both urban and rural areas in Africa where sufficient proportions of larval habitats can be targeted. In this study, we conducted longitudinal larval source surveillance in the western Kenya highlands, generating data which can be used to establish cost-effective targeted intervention tools.MethodsOne hundred and twenty-four (124) positive larval habitats were monitored weekly and sampled for mosquito larvae over the 85-week period from 28 July 2009 to 3 March 2011. Two villages in the western Kenya highlands, Mbale and Iguhu, were included in the study.After preliminary sampling, habitats were classified into four types: hoof prints (n = 21; 17 % of total), swamps (n = 32; 26 %), abandoned goldmines (n = 35; 28 %) and drainage ditches (n = 36; 29 %). Positive habitats occurred in two land use types: farmland (66) and pasture (58). No positive larval habitats occurred in shrub land or forest.ResultsA total of 46,846 larvae were sampled, of which 44.1 % (20,907) were from abandoned goldmines, 30.9 % (14,469) from drainage ditches, 22.4 % (10,499) from swamps and 2.1 % (971) from hoof prints. In terms of land use types, 57.2 % (26,799) of the sampled larvae were from pasture and 42.8 % (20,047) were from farmland. Of the specimens identified morphologically, 24,583 (52.5 %) were Anopheles gambiae s.l., 11,901 (25.4 %) were Culex quinquefasciatus, 5628 (12 %) were An. funestus s.l. and 4734 (10.1 %) were other anopheline species (An. coustani, An. squamosus, An. ziemanni or An. implexus). Malaria vector dynamics varied seasonally, with An.gambiae s.s. dominating during wet season and An.arabiensis during dry season. An increased proportion of An. arabiensis was observed compared to previous studies.ConclusionThese results suggest that long-term monitoring of larval habitats can establish effective surveillance systems and tools. Additionally, the results suggest that larval control is most effective in the dry season due to habitat restriction, with abandoned goldmines, drainage ditches and swamps being the best habitats to target. Both farmland and pasture should be targeted for effective larval control. An increased proportion of An. arabiensis in the An. gambiae complex was noticed in this study for the very first time in the western Kenya highlands; hence, further control tools should be in place for effective control of An. arabiensis.