The ketogenic diet has been used in the treatment of epilepsy for almost a century. Popularity and use has been variable over the years, with the requirement of intense dietetic support. The argument against wide acceptance has often been that there is a lack of appropriate efficacy data, although this point has recently been addressed (Neal et al., 2008). The study within this issue of Epilepsia is to be commended (Freeman et al., 2008). Although a randomized controlled trial of the effect of the ketogenic diet has been performed (Neal et al., 2008), blinding family and professionals to the treatment was thought to be impossible. The study reported herein attempts to address this difficulty, and demonstrates that the methodology is possible. The study also targets a specific seizure type rather than addressing all seizure types, and uses electroencephalography (EEG) to determine the frequency of electrical events. Difficulties in recruitment, however, are highlighted by the duration of the study. In addition, although the clinical efficacy does not quite reach statistical significance, retrospectively, the methodology may have hindered true assessment of effect. Baseline monitoring of seizures occurred during the fasting period—it is difficult to know the initial effect of acquired ketosis (especially as the arm where saccharin was given first appeared to create greater efficacy). Ketosis was also not broken, even with the introduction of carbohydrate. Furthermore, we have no information about the effect on other seizure types, or indeed background EEG activity. The evaluation of seizure frequency for determining treatment response has always been a difficult topic, especially in children. Concern about over- or underreporting may be expressed. What exactly is seizure frequency? Are all events so reported by parents epileptic in origin? Are all events witnessed? The clinical monitoring of absence attacks, in particular, is fraught with difficulty, not least in view of the impossibility of counting all the absence events that manifest. Are they reliably epileptic seizures? Behavioral arrest in this group of children, in particular, may be nonepileptic. Some studies have not used these events as an outcome variable, probably for this reason of interpretation (Motte et al., 1998). Hence, the current use of EEG in many studies, which may add inconvenience and expense. In addition, there will be variability as to what parents see as the most important seizure type. The relevance of subclinical activity in the day-to-day management of children with epilepsy remains a topic of much debate; many parents whose children undertake the ketogenic diet report improved alertness without necessarily a change in seizures (Neal et al., 2005). The underlying mechanisms responsible for such changes remain unclear. One could question the relevance of electrical events if not clinically identified. Therefore, the exact relevance of clinically undetected electrodecremental events remains in question. Interestingly, this study reports more effect of the ketogenic diet on clinical than on EEG events. Conversely, whereas the detection of EEG changes may provide more accurate determination of effect; clinical day-to-day practice encourages treatment of clinical rather than subclinical (e.g., EEG) events in the majority of children.