To assess the causal association of autism with measles, mumps, and rubella (MMR) vaccine versus that with monovalent measles, mumps, and rubella immunization, a 1:2 sex-adjusted logistic regression analysis was conducted using data on subjects who were growing up in the Tokyo area between 1988 and 1992. When MMR immunization was used as a reference, monovalent measles immunization (odds ratio [OR] = 5.33, 99% confidence interval [CI]: 1.03-27.74), non-mumps immunization (OR = 8, 99%CI: 1.33-48.2), and non-rubella immunization (OR = 8.57, 99%CI: 1.30-56.4) with development of autistic spectrum disorders (ASD) were significantly increased. These results suggest a decreased risk of developing ASD with MMR compared to monovalent antigens. However, our findings may reflect potential selection bias due to requiring written consent, possible delayed vaccination in suspected autism cases, and small sample size (case = 21). For the case group and the control group, immunization completeness rate of each antigen, regardless of the timing of immunization, was 90.5% versus 100% in measles, 42.9% versus 78.6% in mumps (P < 0.01), 52.3% versus 83.3% in rubella (P < 0.01), 14.3% versus 45.2% in varicella (P < 0.01), 100% versus 90.5% in polio>2, 100% versus 97.6% in Diphtheria (D), pertussis, and tetanus (T)>3, 85.7% versus 66.7% in DT, 95.2% versus 92.9% in BCG, and 52.4% versus 81.0% in Japanese encephalitis>3 (P<0.01). Only two case subjects and four control subjects received their measles, mumps, and rubella immunizations separately, suggesting that few Japanese parents might have had concerns about the safety of MMR vaccine. A nation-wide study would be a practical measure to scientifically judge the safety of MMR and other routine childhood immunizations.