To present results of optimising scoliosis examination by changing from a conventional film/grid (F/G) to air-gap technique using computed radiography (CR), and to evaluate different methods for estimating effective radiation doses. Forty-nine children and adolescents were examined with an F/G technique, and 21 with air-gap and CR techniques. Entrance surface doses (ESD) were determined with lithium fluoride thermoluminescence dosimeters. For all patients, the effective radiation doses were determined using a hermaphrodite PCXMC computer program. For all F/G radiographs, the effective doses were also determined according to the NRPB-R279 report, and for 22 children (>9 years and/or >40 kg) also with the ODS-60 program, which allows separate gender calculations. Accumulated doses for 37 children examined more than once with F/G examinations were assessed. For F/G techniques, the ESDs for both frontal and lateral views varied with age and were significantly correlated to the patients' thickness. The calculated effective doses using the PCXMC program and the NRPB-R279 did not differ significantly for all frontal and lateral radiographs, respectively, but ODS-60 gave significantly higher values in female subjects. With air-gap and CR techniques, the mean effective doses were reduced by a factor over 10. The mean accumulated effective dose for 37 children with a mean of seven F/G examinations was 6.1 mSv, implying a risk of death of about 1:2,000 for boys and at least 1:1,000 for girls. Paediatric scoliosis radiography should be considered a specialised procedure, which has to be optimised using a non-grid technique.