The concept of the time factor in radiotherapy originated in the controversy surrounding single-dose and fractionated treatments during the first 20 years of this century. The success of Coutard's fractionated treatments of larynx tumors was an important factor in the abandonment of single-dose treatments. There was considerable research afterwards into the influence of dose rate and overall time of treatment on the responses of normal tissues. Recovery was modeled in terms of the Schwarzschild law of photochemistry, as exemplified by the analysis of Strandqvist in log dose-log time coordinates. Different conventions were followed in defining the time for a single-dose treatment. Subsequently the concept arose that the slopes of isoeffect lines relating dose and treatment time for normal tissues and tumors were different and moreover that the effects of fraction number and overall time could be separated; these developments constituted the foundation of the Ellis NSD model. It had an important influence on clinical practice and was reasonably successful in predicting isoeffective regimens for acute effects. It failed to predict severe late effects after large dose fractions. The dissociation between acute and late effects with altered fractionation led to recognition of the importance of the ratio alpha/beta in characterizing the fractionation sensitivity of tissues.