There is a growing body of research on the problems older and disabled people face with authentication systems and a range of solutions have been developed. However, this research has not been integrated across user groups and solutions usually only target one or two groups. This research has attempted to integrate the empirical research findings across studies conducted with people with visual, physical, and intellectual disabilities, people dyslexia and older people. It proposes an initial set of 15 recommendations for the universal design of authentication systems. Four recommendations are about authentication systems in general, six about password authentication, two about CAPTCHAs and three about biometric authentication. 40% of the recommendations address problems experienced by at least two different user groups and over 25% address problems experienced by three groups. However, much further work is needed to validate and refine the recommendations and integrate them into a universal design approach.