We review published epidemiologic studies on personal use of hair dyes and leukemias and myelodysplastic syndromes (MDS). A subsequent article will review studies on lymphomas and multiple myeloma. A computerized literature search for the years 1966 through 1996 was completed using MEDLINE. Data were extracted using a standardized form that recorded study design, study population, type of cases, comparison group, sources of data on personal exposure to hair dyes, method of data collection, type of exposure data collected, covariates, and results. The above search identified 13 epidemiologic studies on the possible association between personal use of hair dyes and leukemias and MDS. Although there are some reports of positive associations, overall the evidence linking personal use of hair dyes to various leukemia and MDS subgroups is weak. One cannot definitively rule out an association, however, because of the methodologic limitations, such as small numbers of exposed cases and lack of detailed exposure information. Any further research would need much better assessment of hair dye use, including product type, color frequency, duration, and changes in use over time, and adequate statistical power.