Research has shown that phonological decoding is critical for orthographic learning of new words during independent reading. Moreover, correlational studies have demonstrated that the strength of orthographic learning is related to the orthographic knowledge with which readers approach a text. The present training study was conducted to assess experimentally whether this relation between prior orthographic knowledge and orthographic learning while reading is causal by assessing whether instruction designed to increase sublexical orthographic knowledge would facilitate orthographic learning during independent reading. A group of Danish-speaking third graders (n = 21) was taught conditional spelling patterns conforming to the opaque Danish writing system, with emphasis on how to map the spellings onto their pronunciations. A matched control group (n = 21) received no treatment. Both groups were exposed to 12 novel words containing trained spelling patterns in an orthographic learning task. Posttests revealed a moderate transfer effect from training to orthographic learning, measured as the students’ ability to identify target word spellings in an orthographic choice task, and a strong transfer effect when measured as their ability to reproduce target word spellings in a spelling task. However, no advantage of explicit training over reading only could be detected when orthographic learning was measured as target word naming. The findings support the view that larger sound spelling units are used to form connections between spellings and pronunciations of words. Additionally, the findings support the view that preexisting orthographic knowledge is causally related to the degree and quality of orthographic learning during independent reading.