The effects of Content and Language Integrated Learning (CLIL) on students' foreign language skill development have been overestimated by previous studies, as most studies needed to have considered selection and preparation effects appropriately. We used complete survey data from a 1996-2003 cohort to investigate English skill development of N = 332 German CLIL and N = 6,401 non-CLIL grammar school students from grade seven (M = 12.64, SD = 0.60) to grade eleven (M = 16.72, SD = 0.70). We found selection effects for prior achievement, sociodemographic variables, and cognitive abilities. After propensity score matching, data revealed significant preparation effects of additional English lessons for the CLIL students. When controlling selection and preparation effects, CLIL compensated for the assumed fading out-effect but did not contribute significant added value measured by a C-test. We commend to include selection and preparation effects when analyzing CLIL effects.