About 90% of the 0.5 million global annual death toll from malaria occurs in Sub-Saharan Africa, especially amongst young children. While the most popular interventions for reducing the burden of malaria include insecticide-treated nets, improved dwellings, spraying and modification of mosquito habitat (e.g. drainage), use of some of these interventions has reached a plateau in some countries. Recently, a new anti-malaria intervention of reducing deforestation is claimed to be effective. The pathways through which deforestation can increase malaria include ecological effects that increase efficiency of the mosquito vectors and socio-economic changes such as migration that expose people with little tolerance to endemic forms of malaria. It is therefore argued in some studies that reducing deforestation can reduce the prevalence of malaria, presenting a win-win for health and the environment. The evidence on the effect of deforestation on malaria comes especially from linking remote sensing data on forest change to survey reports made by mothers of whether their child had a fever. For example, one influential study from Nigeria estimated the prevalence of malaria by asking mothers whether their child had been ill with a fever during the two weeks preceding the survey. Fever in children can have many causes other than malaria and so this proxy variable may lead to misleading relationships that divert attention from more effective anti-malaria interventions. In this study we use Demographic and Health Survey (DHS) data from Nigeria and Tanzania which has results from Rapid Diagnostic Tests (RDT) for parasite antigens in a blood drop, and tests using microscopic examination for Plasmodium in blood slides. We contrast the effects of local deforestation on these measured forms of malaria with the effects of deforestation on mother's report of children's fever (coming from the same DHS samples). In contrast to some recent studies, a positive impact of deforestation on malaria prevalence is not apparent when measured malaria (using microscopy or RDT) is the outcome variable. However, effects of deforestation on fever in children are apparent. The contrast between the impact on a measured health outcome and the impact on a proxy measure (the mother's report of fever in children) suggests that caution is needed when using proxy measures of health outcomes. In terms of policy implications, despite the potential attractiveness of adding child health reasons to the list of reasons for reducing the rate of deforestation, the available evidence for these two sub-Saharan African countries does not support this interpretation, in contrast to some recent studies.