Summary.Two experiments were carried out to test the hypothesis that different methods and different media in the teaching of reading give rise to different kinds of spelling error. Comparisons were made between children taught to read by look‐and‐say and by phonic methods, and between children taught in traditional orthography (t.o.) and the initial teaching alphabet (i.t.a.). When the methods are ranked for vulnerability to particular types of error, a trend is revealed in which i.t.a. produces the fewest omissions, insertions and perseverations, the look‐and‐say method produces the fewest doubling errors, consonant substitutions, faulty auditory perceptions and homophones. Between these two extremes, lies the phonic method which gives fewest transpositions, vowel substitutions and the greatest number of reasonable phonic alternatives. Differences between the look‐and‐say and phonic methods are fully consistent with the differential emphasis placed upon their ‘perceptual’ and ‘rule‐following’ aspects. Differences between t.o. and i.t.a. are consistent with differences in economy of form. Breakdown of results by sex and I.Q. reveals significant differences in the lower I.Q. range. It is concluded that though overall spelling attainment is not affected, perceptual and rule‐following skills influence spelling in a way which may have important implication