This study investigates whether learning tests are able to assess the chang
e potential of intellectual ability better than traditional intelligence ap
proaches. The learning test approach is compared to competing diagnostic ap
proaches (traditional status diagnostics, diagnostics of domain-specific pr
ior-knowledge). Further, it is shown that learning tests make a substantial
contribution to a long-term prediction of school learning performance. The
article reports two validation studies with 166 pupils of the 7th grade (c
oncurrent validation) and 130 pupils of the 8th grade level (prognostic val
idation). Learning performances in computer-assisted, standardized and curr
icular-related tutorials were used as external validity criteria. The resul
ts show that learning tests are able to predict future learning performance
. It can be shown that learning tests explain criterion variance independen
tly from measures of prior-knowledge and of intelligence status (measured b
y traditional intelligence tests). Learning tests provide additional inform
ation to diagnostician which cannot be supplied by conventional status-orie
nted procedures.