Benzene is considered a carcinogen, mostly based on evidence of causality for myeloid leukemia from high levels of exposure in occupational studies. We used United States Environmental Protection Agency National Ambient Toxics Assessment (NATA) estimates of low‐level ambient benzene to examine potential associations for the general public between benzene exposure and risk of hematologic cancers. Exposure was estimated by linking participants’ residential address to the NATA benzene estimates for that census tract. Among 115,996 American Cancer Society Cancer Prevention Study‐II Nutrition cohort participants (52,554 men, 63,442 women), 2,595 were diagnosed with incident hematologic cancer between 1997 and 2013. Extended Cox regression modeling was used to estimate hazard ratios (HR) and 95% confidence intervals (CI). Among all participants, ambient benzene was positively associated with myelodysplastic syndromes (HR = 1.16, 95% CI: 1.01–1.33 per μg/m3) and T‐cell lymphoma (HR = 1.29, 95% CI: 1.08–1.53 per μg/m3). Among men, ambient benzene was also positively associated with any hematologic malignancy (HR = 1.07, 95% CI: 1.01–1.15 per μg/m3) and follicular lymphoma (HR = 1.28, 95% CI: 1.09–1.50 per μg/m3). No significant associations were observed for women only, but associations were suggestive for MDS and T‐cell lymphoma. It is possible that the NATA ambient benzene estimates are a better proxy for benzene exposure for men than women in this cohort. The results of this study support an association between ambient benzene and risk of hematologic malignancies, particularly MDS, T‐cell lymphoma and follicular lymphoma. More research in large scale or pooled studies is needed to further explore these associations.