Repellents prevent mosquito bites and help reduce mosquito-borne disease, a global public health issue. Laboratory-based repellent bioassays predict the ability of compounds to deter mosquito feeding, but the variety of repellent bioassays and statistical analysis methods makes it difficult to compare results across methodologies. The most realistic data are collected when repellents are applied on the skin; however, this method exposes volunteers to chemicals and mosquito bites. Silicone membranes were investigated as an alternative to human skin in assays of repellent efficacy. Results from module system bioassays conducted in vitro with a silicone membrane were compared with in vivo bioassays conducted with N,N-diethyl-3-methylbenzamide (referred to as DEET), 1-piperidinecarboxylic acid 2-(2-hydroxyethyl)-1-methylpropylester (referred to as Picaridin), ethyl 3-[acetyl(butyl)amino]propanoate (referred to as IR3535), and para-menthane-3,8-diol (referred to as PMD) applied directly on the skin of the leg. No significant difference in mosquito feeding was found when comparing skin and volunteer-worn membrane controls using blood; however, feeding was significantly lower in unworn membrane controls using either 10% sucrose or blood, indicating that worn membranes are a possible surrogate for untreated human skin. Pooled data from six volunteers were used to generate dose–response curves of blood-feeding activity. Results from skin-applied repellents were modeled to determine if membranes could provide a predictive correlate for skin. Goodness-of-fit comparisons indicated that the nonlinear dose–response curves for the skin and membrane differed significantly for DEET and Picaridin, but did not differ significantly for IR3535 and PMD. With knowledge of the dose–response relationships and further modifications to this system, the membrane-based tests could be used for standardized repellent testing with infected vectors