In this full research-to-practice paper we study novice programming students learning process in the computer laboratory. Working with laboratory assignments is an important component when students learn to program. Here the assignments are intended to help students consolidate theoretical understanding and simultaneously train practice. However, it has been observed that the learning outcome of such laboratory sessions often is unsatisfactory. In this article we ask the question "How do novice students go about learning in the computer laboratory?" We analyse empirical data on novice students working in pairs in the laboratory, which is common in a first programming course. The data consists of video films of students where they discuss and solve programming problems, screen captures of what the students typed during the same laboratory session, and stimulated recall interviews with the students after the laboratory session. In the analysis we use an approach inspired by phenomenography and variation theory. We specifically focus on typical stages in the learning process when students learn in the programming laboratory. In doing so we have identified successful and less successful learning paths, where variation can play different roles. The stages identified in students’ learning process are I. Students first need to become aware of a lack of clarity. In the data we have identified different ways in which this necessary awareness was trigged; II. If, and in that case how, they resolve the lack of clarity. In all the stages we found successful and less successful ways in which students’ handle the situations. We illustrate the stages and discuss how and why variation may play different roles in the different stages of students’ learning, specifically focusing of the unsuccessful learning paths. Lastly, we discuss what these findings can tell us about how programming labs could be designed to promote learning in terms of helping students to avoid the unsuccessful paths identified.