The aim of the present study was to examine whether reading or writing self-efficacy rather than actual skills could predict participation in prison education when controlled for age, sentence length, and education level. Six hundred subjects from a representative sample of prisoners in Norway completed a questionnaire, and a subsample of 92 of these subjects also completed a standardized reading and spelling test. A logistic regression analysis indicated that higher efficacy beliefs with regard to writing increased the likeliness of participating in prison education, while actual reading or spelling skills did not predict participation. Younger prisoners were more likely participate compared to older ones. Prisoners with longer sentences, were twice as likely to participate in prison education as those with shorter sentences, and those who self-reported a diagnosis of specific reading and spelling difficulties/dyslexia were twice as likely to participate and those not reporting such difficulties. This is in line with former studies in countries with relatively short sentences.