It is unclear whether early psychosis in the context of cannabis use is different from psychosis without cannabis. We investigated this issue by examining whether abnormalities in oculomotor control differ between patients with psychosis with and without a history of cannabis use. We studied four groups: patients in the early phase of psychosis with a history of cannabis use (EPC; n = 28); patients in the early phase of psychosis without (EPNC; n = 25); controls with a history of cannabis use (HCC; n = 16); and controls without (HCNC; n = 22). We studied smooth pursuit eye movements using a stimulus with sinusoidal waveform at three target frequencies (0.2, 0.4 and 0.6 Hz). Participants also performed 40 antisaccade trials. There were no differences between the EPC and EPNC groups in diagnosis, symptom severity or level of functioning. We found evidence for a cannabis effect ( χ 2 = 23.14, p < 0.001), patient effect ( χ 2 = 4.84, p = 0.028) and patient × cannabis effect ( χ 2 = 4.20, p = 0.04) for smooth pursuit velocity gain. There was a large difference between EPC and EPNC ( g = 0.76–0.86) with impairment in the non cannabis using group. We found no significant effect for antisaccade error whereas patients had fewer valid trials compared to controls. These data indicate that impairment of smooth pursuit in psychosis is more severe in patients without a history of cannabis use. This is consistent with the notion that the severity of neurobiological alterations in psychosis is lower in patients whose illness developed in the context of cannabis use.