Objectives Contemporary fear-avoidance models of chronic pain posit that fear of pain, and overgeneralization of fear to non-threatening stimuli is a potential pathway to chronic pain. While increasing experimental evidence supports this hypothesis, a comprehensive investigation requires testing in multiple modalities due to the diversity of symptomatology among individuals with chronic pain. In the present study we used an established tactile fear conditioning paradigm as an experimental model of allodynia and spontaneous pain fluctuations, to investigate whether stimulus generalization occurs resulting in fear of touch spreading to new locations. Methods In our paradigm, innocuous touch is presented either paired (predictable context) or unpaired (unpredictable context) with a painful electrocutaneous stimulus (pain-US). In the predictable context, vibrotactile stimulation to the index or little finger was paired with the pain-US (CS+), whilst stimulation of the other finger was never paired with pain (CS-). In the unpredictable context, vibrotactile stimulation to the index and little fingers of the opposite hand (CS1 and CS2) was unpaired with pain, but pain-USs occurred unpredictable during the intertrial interval. During the subsequent generalization phase, we tested the spreading of conditioned responses (self-reported fear of touch and pain expectancy) to the (middle and ring) fingers between the CS+ and CS-, and between the CS1 and CS2. Results Differential fear acquisition was evident in the predictable context from increased self-reported pain expectancy and self-reported fear for the CS + compared to the CS-. However, expectancy and fear ratings to the novel generalization stimuli (GS+ and GS-) were comparable to the responses elicited by the CS-. Participants reported equal levels of pain expectancy and fear to the CS1 and CS2 in the unpredictable context. However, the acquired fear did not spread in this context either: participants reported less pain expectancy and fear to the GS1 and GS2 than to the CS1 and CS2. As in our previous study, we did not observe differential acquisition in the startle responses. Conclusions Whilst our findings for the acquisition of fear of touch replicate the results from our previous study (Biggs et al., 2017), there was no evidence of fear generalization. We discuss the limitations of the present study, with a primary focus on procedural issues that were further investigated with post-hoc analyses, concluding that the present results do not show support for the hypothesis that stimulus generalization underlies spreading of fear of touch to new locations, and discuss how this may be the consequence of a context change that prevented transfer of acquisition.