The advent of extensive small molecule databases has brought with it the problem of searching these repositories for entities with desired properties. A multitude of similarity-searching algorithms, based on different underlying methods, currently exist for this purpose. However, due to the somewhat nebulous definition of "similar", all such approaches maintain different strengths and weaknesses. Presented here is PZIM, a new approach fundamentally based on a description of the atom environment that includes multiple adjustable features allowing for searches to be tailored on the basis of the user definition of similarity. In addition to flexible atom environment size, PZIM allows for the use of an atom-substitution matrix to identify similar pharmacophoric recognition elements. Finally, PZIM produces 2-dimensional alignments of all compared molecules that pass a user-defined threshold for similarity. To determine the usefulness of the approach, PZIM was compared to seven other similarity-searching methods on nine data sets recently published. PZIM achieved a rank of first or second in the majority of cases tested and obtained the highest average rank score across all methods tested. These results demonstrate the effectiveness of the PZIM approach across diverse search conditions.