This study compares scores in eductive ability, measured by Raven’s Standard Progressive Matrices, in three groups of 8th graders, examined in 1988, 1993, and 1998. It was expected to find a significant gain in raw scores in each examination moment-known as Flynn effect-, as has been verified in several industrialized countries. Surprisingly, in opposition to this hypothesis, a significant decrement in raw scores was found between 1993 and 1998, with no association with the rest of the variables included in the design (sex, parental educational and occupational level, cohabitation with one or both parents, sibling existence and class schedule). Results are discussed focusing in environmental variables.Se presenta un estudio cross-secconal, comparativo de desempeños en capacidad educativa, medidos por el Test de Matrices Progresivas, Escala General, en tres grupos de adolescentes de Buenos Aires, Argentina (N=571), que asistían al 8° año de la EGB en 1988, 1993 y 1998. Se hipotetizaba comprobar el efecto Flynn, consistente en un incremento estadísticamente significativo de las puntuaciones en cada momento de examen, ya verificado en varias naciones industrializadas. Contrariamente, los resultados mostraron un decremento significativo entre 1993 y 1998, no asociado con ninguna de las variables consideradas (sexo, educación y ocupación de los adultos a cargo, convivencia con uno o ambos padres, existencia de hermanos y turno escolar). Los resultados se discuten considerando la posible influencia de variables ambientales.