Retention of Spanish learned in school was tested over a 50-year period for 733 individuals. Tests of reading comprehension, recall, and recognition vocabulary and grammar were administered together with a questionnaire to determine the level of original training, the grades received, and rehearsals during the retention interval in the form of reading, writing, speaking, or listening to Spanish. Multiple regression analysis shows that retention throughout the 50-year period is predictable on the basis of the level of original training. The great majority of subjects rehearse so little that the data reveal no significant rehearsal effects. The analysis yields memory curves which decline exponentially for the first 3-6 years of the retention interval. After that retention remains unchanged for periods of up to 30 years before showing a final decline. Large portions of the originally acquired information remain accessible for over 50 years in spite of the fact the information is not used or rehearsed. This portion of the information in a "permastore" state is a function of the level of original training, the grades received in Spanish courses, and the method of testing (recall vs. recognition), but it appears to be unaffected by ordinary conditions of interference. The life-span frequency distribution of learned responses is discontinuous; one portion of the response distribution has life spans of 0-6 years, the other portion, life spans in excess of 25 years, and no responses have life spans of 6-25 years. This suggests a discrete transition into a permastore state during the extended period of original training. Analysis of successive relearning processes over extended time periods is deemed essential for an understanding of the acquisition of permanent semantic memory content.