In this human fear conditioning study, the online development of conditioned US-expectancy to discrete cues and background contexts was measured in two groups. In the paired group (n=30), the CS was systematically followed by an aversive shock (US). In the unpaired group (n=30), CS and US were presented explicitly unpaired. Using US-expectancy ratings, we replicated the basic finding already illustrated in humans with startle modulation. In the paired group, the CS elicited more US-expectancy than the context, whereas in the unpaired group, the context elicited more US-expectancy than the CS. Interestingly, we also observed a trial-by-trial development of conditioning to the context in the unpaired group as indicated by a significant linear trend. This gradual development and the evidence for the role of US-expectancy in contextual fear add to the idea that cued and contextual fear rely on the same basic associative processes.