One hundred and sixty-four accessions representing Czech and Slovak pea (Pisum sativum L.) varieties bred over the last 50 years were evaluated for genetic diversity using morphological, simple sequence repeat (SSR) and retrotransposon-based insertion polymorphism (RBIP) markers. Polymorphic information content (PIC) values of 10 SSR loci and 31 RBIP markers were on average high at 0.89 and 0.73, respectively. The silhouette method after the Ward clustering produced the most probable cluster estimate, identifying nine clusters from molecular data and five to seven clusters from morphological characters. Principal component analysis of nine qualitative and eight quantitative morphological parameters explain over 90 and 93% of total variability, respectively, in the first three axes. Multidimensional scaling of molecular data revealed a continuous structure for the set. To enable integration and evaluation of all data types, a Bayesian method for clustering was applied. Three clusters identified using morphology data, with clear separation of fodder, dry seed and afila types, were resolved by DNA data into 17, 12 and five sub-clusters, respectively. A core collection of 34 samples was derived from the complete collection by BAPS Bayesian analysis. Values for average gene diversity and allelic richness for molecular marker loci and diversity indexes of phenotypic data were found to be similar between the two collections, showing that this is a useful approach for representative core selection.