The paper discusses the procedures in building of CroDeriV, the computational lexicon containing data on morphological structure of Croatian verbs. CroDeriV is the first morphological resource dealing with derivational phenomena of Croatian. In the first part of the paper, after the motivation for building this kind of lexicon and the brief overview of the existing morphological resources for Croatian, procedures for morphological segmentation of verbs in CroDeriV are presented. Each verb in CroDeriV is segmented into lexical and derivational morphemes. Verbs of the same root are mutually linked. This procedure enables the recognition of derivationally related families of verbs and, at the same time, the detection of full derivational spans of particular base forms. The second part of the paper focuses on the morphological structure of Croatian verbs based on the analysis of almost 14000 verbal lemmas currently included in CroDeriV. The analysis enabled the recognition of generalized morphological structure applicable to all Croatian verbs. It consists of four slots for derivational prefixes and three slots for derivational suffixes on each side of a lexical morpheme, and these slots are provided for every verbal lemma in CroDeriV. This structure is compared with other approaches dealing with morphology of Croatian verbs. The three suffixal slots and their semantics are explained in more detail, since this kind of segmentation has so far not been introduced in Croatian morphological literature. First suffixal slot comprises suffixes with specialized meanings (eg diminutive, pejorative), second slot suffixes with aspectual meaning, and third slot suffixes denoting conjugational class. The final part of the paper describes attested combinations of derivational affixes in CroDeriV and indicates the frequency of their occurrence.