Background Benzene exposure has been associated with increased risk of leukaemia and myelodysplastic syndrome. Existing studies are sparse for other lymphohaematopoietic cancer subtypes, such as myeloproliferative disease (MPD) and the related chronic myeloid leukaemia (CML). We pooled data from three petroleum worker nested case?control studies to address this gap. To our knowledge, this is the first study to systematically examine the relationship between MPD and quantitative benzene exposure. Methods There were 28 cases and 122 matched controls for CML and 30 MPD cases with 124 matched controls. Two haematopathologists identified each case and provided a diagnosis certainty score. Blinded datadriven assessments estimated benzene exposure for each job held by study participants. Statistical analyses included conditional logistic regression and penalised smoothing splines. Results Benzene exposures were low, and mean average exposure intensity for CML cases was 0.3 ppm and for MPD cases 0.17 ppm. Categorical analyses showed no increased risk of CML or MPD with benzene exposure. There was no significantly increased risk identified for more highly exposed terminal workers. Some association was seen in spline analyses between increased risk of MPD and benzene exposure experienced in the 2?20 years before diagnosis and with peak exposures considered with cumulative exposure as a continuous variable. Conclusions No convincing association was identified between MPD or CML and low exposure to benzene. The greater risk for exposures experienced in the 20 years before diagnosis needs investigating in more powerful studies with a wider range of exposure to benzene, and the biological plausibility further examined from a mechanistic viewpoint.