The design of applications using mobile devices needs a different quality assessment than those known for desktop applications. Of the many aspects that have to be taken into account, one important criterion is the average time users need to complete a task. For interactions with the mouse, keyboard or touch screens, there exist models that predict interaction times like Fitts' law or the Keystroke-Level Model (KLM). This paper shows parallels to these models for advanced interactions with mobile phones targeted at pervasive services, including near field communication as well as built-in cameras and sensors. Applications can be evaluated with respect to user performance time without having a prototype running on the phone. To accomplish that, we extend the known KLM by identifying basic interaction elements for mobile phones and give estimates for expert user performance derived from several user tests.