Abstract Background This Scandinavian collaborative retrospective study of children treated with ketogenic diet (KD) highlights indications and effectiveness over two years follow-up. Methods Five centres specialised in KD collected data retrospectively on 315 patients started on KD from 1999 to 2009. Twenty-five patients who stopped the diet within four weeks because of compliance-problems and minor side-effects were excluded. Seizure-type(s), seizure-frequency, anti-epileptic drugs and other treatments, mental retardation, autism-spectrum disorder and motor-dysfunction were identified and treatment-response was evaluated. Results An intention-to-treat analysis was used. Responders (>50% seizure-frequency reduction) at 6,12 and 24 months were 50%, 46% and 28% respectively, seizure-free were 16%, 13% and 10%. Still on the diet were 80%, 64% and 41% after 6, 12 and 24 months. No child had an increased seizure-frequency. The best seizure outcome was seen in the group with not-daily seizures at baseline (n = 22), where 45%, 41% and 32% became seizure-free at 6, 12 and 24 months A significant improvement in seizure-frequency was seen in atonic seizures at three months and secondary generalised seizures at three and six months. Side-effects were noted in 29 subjects; most could be treated and only two stopped due to hyperlipidaemia and two due to kidney-stones. In 167 patients treated with potassium-citrate, one developed kidney-stones, compared with six of 123 without potassium-citrate treatment (relative risk = 8.1). Conclusions As the first study of implementing KD in children in the Scandinavian countries, our survey of 290 children showed that KD is effective and well tolerated, even in such severe patients with therapy-resistant epilepsy, more than daily seizures and intellectual disability in the majority of patients. Long-term efficacy of KD was comparable or even better than reported in newer AEDs. Addition of potassium citrate reduced risk of kidney-stones. Our data indicate that the response might be predicted by seizure-frequency before initiation of the diet but not by age, seizure-type or aetiology.