We report and contrast the results of a series of priming studies in Italian and Spanish, whose main goal was to empirically verify the psychological reality of two features crucially involved in event type classification: resultativity and durativity. Unlike most experimental research, focused on ontological properties of actional classes of predicates (Todorova et al. 2000, Gennari & Poeppel 2003, Pylkkänen & McElree 2006, Bott 2010), we adopt a feature-based approach to verbal semantics. Semantic priming is one of the most common experimental paradigms to probe the organization of the lexicon (McNamara 2005). We apply it here to get behavioral data about the role of event type features of verb meaning. We followed the general design in Bonnotte (2008), who tested French achievement and activity verbs (respectively classified as [+resultative, -durative] and [-resultative, +durative]) in two semantic decision tasks: the subjects were asked if the target referred to a durable or non-durable situation and whether it implied a clear outcome or not. Our Italian and Spanish semantic decision experiments presented a number of modifications in data selection with respect to Bonnotte’s design. First of all, we ran an event classification pre-test to make sure that only the most unambiguously identified achievements and activities were included in the data set. Further modifications were introduced to minimize any semantic effect not related to event types (the semantic relatedness of primes and targets was measured in a web-based pre-test), and to protect the actional classes tested from any factor that might blur the semantic interpretation (punctual verbs, which are often confused with achievements, were excluded from the data set). A total of 36 verbs (18 activities and 18 achievements) were selected for the final data set. The pattern of priming effects obtained for the Romance languages presents some striking similarities: in the resultativity task, only achievements benefited from priming from opposite primes. There are, however, some intriguing differences, too: in the durativity task, activities were processed significantly faster after similar primes in Italian but in Spanish only achievements were primed. In addition to the semantic decision task, we ran a lexical decision task to check whether both features impact the processing even when the speakers do not have to consciously identify the verbs marked with these features. In other words, we wanted to check whether these features exert their effect in an on-line task, in an automatic and implicit way. The results obtained for Spanish seem to confirm that this is indeed the case: achievements benefited from opposite primes and activities from similar primes. This replicates exactly the pattern observed in the semantic decision task for Italian and, in part, for Spanish. Data collection for Italian verbs with a similar lexical decision task is currently ongoing. This seems to suggest that the durativity and resultativity features can be selectively activated in the mental representation of verbal semantics, bringing further empirical support to the feature-based approach to event types.