Various metamodeling approaches are emerged as powerful means to alleviate computational challenge of Monte Carlo simulation (MCS) based reliability analysis of structure involving implicit limit state function. In this regard, support vector regression (SVR) based metamodel has revealed superior performance using small sample learning. The present study explores the effectiveness of an adaptive SVR based metamodeling approach for structural reliability analysis (SRA). A two-stage adaptive scheme is proposed to effectively obtain the necessary training data as close as possible to the limit state to construct SVR based metamodel for SRA. In the first stage, an initial design of experiment is built by a space-filling design over the entire physical domain of the random variables to construct an SVR model. In the next stage, based on the prediction at MCS points using the previous SVR model, a subset of MCS samples are selected. These are now used to enrich the existing training data by adding data points sequentially such that new points are closest to the limit state and as far as possible from the existing points. The effectiveness of the proposed approach is elucidated numerically. The results show much improved estimate of reliability by the proposed adaptive SVR based metamodeling scheme considering the direct MCS based reliability results as the benchmark.