As a cyclist, it is essential to make inferences about the intentions of other road users in order to anticipate their behaviour. There are official ways for cyclists to communicate their intentions to other road users, such as using their arms to point in the intended direction of travel. However, in everyday traffic cyclists often do not use such active forms of communication. Therefore, other visual cues have to be used to anticipate (critical) encounters or events. During this study, 108 participants completed a video internet survey in which they predicted the intentions of a lead cyclist based on visible behaviour preceding a turning manoeuvre. When the lead cyclist approached the intersection, each video was stopped just before the cyclist initiated turning. Based on visual cues, the participants had to select which direction they thought the cyclist would go. After entering their prediction, they were asked how certain they were about their prediction and on which visible behaviour(s) each prediction was based. The results show that it is very hard to predict the direction of a turning cyclist based on visual cues before the turning manoeuvre is initiated. Exploratory regression analyses revealed that observable behaviours such as head movements and cycling speed were related to prediction accuracy. These results may be used to support cyclists in traffic interactions.