Objective: To evaluate the accuracy of death certificates (DCs) for amyotrophic lateral sclerosis (ALS) in different parts of Italy. Studies based on DC diagnosis for ALS have shown a reduced mortality comparing northern with southern Italy. These data are in contrast with results from other surveys on the incidence of ALS performed in Italy and other countries. Methods: Archives of neurological clinics from northern (Milano, Monza, Pavia, and Bologna) and southern Italy including islands (Napoli, Sassari, Palermo, and Messina) were searched for patients discharged with a diagnosis of ALS in the period 1970–1995. Subjects affected by definite/probable ALS according to the Scottish Motor Neuron Disease Research Group diagnostic criteria were included. DCs were obtained from the vital statistic bureau. True positive rates (TPRs) and 95% confidence intervals (CIs) for proportions were calculated for northern and southern Italy separately. Multiple logistic regression analysis was performed according to gender, age at onset, age and year of death, and interval between onset and death. Results: We found 651 patients affected by definite/probable ALS; 573 of them had died by December 31, 1996. DCs were available for 566 subjects (411 from northern Italy and 155 from southern Italy). TPR was 66.7% (95% CI 61.9–71.2) for northern Italy and 51.6% (95% CI 43.5–59.7) for southern Italy (χ2 = 10.9, p = 0.001). Logistic regression analysis showed an association between a lower accuracy of DCs and the interval between onset of symptoms and death. TPR calculations considering different death periods (1970–1982 and 1983–1996) showed comparable rates of accuracy over time. Conclusions: Mortality statistics based on official death records do not accurately reflect interregional mortality for ALS in Italy.