Acute lymphoblastic leukemia (ALL) in the childhood peak may be a rare response to delayed first exposure to one or more common infectious agent(s). Mycoplasma Pneumonia has the appropriate socioeconomic correlates and clinical symptoms and the hypothesis that delayed first exposure to it may contribute to ALL is considered. Counts of positive reports of M Pneumonia from disease surveillance data for England and Wales (United Kindom) for 1975-92 have been taken as proxies for community burden of infection. Variation by months of birth (cohort) and diagnosis (period) of incidence of ALL in children born and diagnosed 1975-92 are compared with predictions. When periods were classified by mean M Pneumonia count rate in the nine preceding months, standardized morbidity ratios (SMR) for the highest and lowest 20 percent were 108 and 89 (rate ratio [RR] = 1.2, 95 percent confidence interval [CI] = 1.1-1.4). SMRs for cohorts with highest and lowest predicted risk (i.e., lowest and highest M Pneumonia count rate around birth and during infancy) were 110 and 97 (RR = 1.1, CI = 1.0-1.3). The trend for period was most marked in the cohorts with low opportunity for exposure when young. This ecologic analysis provides preliminary support for the hypothesis.