Word associations are among the most common paradigms for studying the human mental lexicon. While their structure and types of associations have been well studied, surprisingly little attention has been given to the individual differences in association behavior and factors affecting such behavior (emotional state and traits of respondents, type of cue words, etc.). Moreover, most research which use word association methodology rely on the assumption of homogeneity of word association behavior and available association norms. Individual association behavior is understudied. To study word association production at the level of individual, datasets are needed which contain data at the individual level, while existing association norms contain data aggregated on the level of a cue word. In this paper we present RuPersWordAssociation dataset which is, to the best of our knowledge, the largest existing (at least from publicly available) word association database with plenty of metadata (demographics, personality traits of respondents, reaction time, semantic similarity metrics) and linguistic annotation for the type of a relation between a cue word and associates (more than 22000 cue – associate pairs were annotated). We provide a detailed description of the dataset and the results of linguistic annotation and present the outcomes of the experiments which revealed individual preferences for certain types of associates. We also highlight some future directions of the application of the presented dataset to study individual mental lexicon and factors affecting individual association behavior.