Light detection and imaging at very low intensities, even down to individual photon detection, has been made possible by the development of photomultiplier and imaging tubes. They have two main components: a photocathode detector and an electron multiplication section, but throughout their long history their performance has been limited by the response of the photocathodes. Reappraisal of the underlying science of cathode preparation emphasizes that they are still performing well below their full potential. The reasons for this are discussed and some indications of how improvements could be made are suggested. Since many of the potential advances are within the scope of current technology it is certainly feasible to achieve enhancements in performance by factors of two to ten across the blue to near-infrared spectral range. Higher improvement factors are implied but will be difficult to realize in the normal spectral range for wavelengths below 1 m. Significant gains in sensitivity, and perhaps in response speed, will open many new opportunities for low light level sensors and the prediction is that they will be an enabling technology for developments in biological and medical applications, among others.