The way users interact with mobile applications varies according to the context where they are. We conducted a study where users had to manipulate a multimodal questionnaire in 4 different contexts (home, park, subway and driving), considering different variables (lighting, noise, position, movement, type of content, number of people surrounding the user and time constraints) that affect interaction. This study aimed at understanding the effect of the context variables in users’ choices regarding the interaction modalities available (voice, gestures, etc). We describe the results of our study, eliciting situations where users adopted specific modalities and the reasons for that. Accordingly, we draw conclusions on users’ preferences regarding interaction modalities on real life contexts.