Current malaria vector control strategies in Africa target indoor resting and biting mosquitoes and rely heavily on a small number of insecticides. These interventions have lead to the selection of insecticide resistance, behavioural adaptations of the vectors and leave naturally exophilic species nearly untouched. Gravid Anopheles gambiae s.l. female searching for an oviposition site would be a novel target for vector control. However, little is known about the oviposition-site selection behaviour (criteria) of this mosquito. The major aim of the presented research was to investigate if gravid An. gambiae s.l make informed choices when selecting an oviposition site and to identify physical, chemical and biological parameters associated with these choices under standardized experimental and natural field conditions. Standardized field tests and dual-choice oviposition bioassays were used to evaluate responses to soil and rabbit food pellets infusions and habitat water and also to test if bacteria and the volatile chemicals that bacteria produce are relevant to habitat selection. A case–control approach was used to study natural aquatic habitats on Rusinga Island in Lake Victoria during the long rainy season in 2012 to compare the characteristics of habitats colonized (cases) and not colonized (controls) by early instar Anopheles larvae. Factors evaluated included biological characteristics of the sites, zooplankton, invertebrate fauna, physical parameters, nutrients, bacteria communities and volatile chemicals released from the water. Multivariate analyses were used to investigate associations between oviposition site characteristics and habitat selection by Anopheles. The experimental work illustrated that wild and caged An. gambiae s.l. females discriminate between potential aquatic habitats for oviposition and gravid An. gambiae s.l. female select suitable habitats using preferred and avoided chemical cues from water bodies. It furthermore emphasizes that natural infusions can be used to manipulate the oviposition behaviour of An. gambiae s.l.. In the field no evidence was found that bacteria from natural habitat water were involved in habitat selection. Although chemical cues were highly diverse analysis suggests that cases and control habitats differ in the headspace volatile profile of the water. High turbidity >200 nephelometric turbidity units (NTU) was the only environmental factor strongly associated with cases. Other risk factors were higher grass coverage (positive association), and the abundance of creeping water bugs of the family Naucoridae and fish (negative associations). This study demonstrates that gravid An. gambiae females choose suitable habitats for oviposition using a complex system of chemical and visual cues from water bodies. Habitats preferred by An. gambiae exhibited distinct and measurable characteristics that can be potentially exploited to attract and kill gravid females to improve malaria vector monitoring and control.