Retrieval is often considered the most important task in Case-Based Reasoning (CBR), since it lays the foundation for overall performance of CBR systems. In CBR, a typical retrieval strategy is realized through similarity knowledge encoded in similarity measures. This strategy is often called similarity-based retrieval (SBR). This paper proposes and validates that association analysis techniques can be used to improve SBR. We propose a retrieval strategy USIMSCAR that performs the retrieval task by integrating similarity and association knowledge. We show its reliability, in comparison with several retrieval methods implementing SBR, using datasets from UCI ML Repository