In ‘On Beginning Treatment’ I compare and contrast the ‘correct’ procedure for beginning an analysis that I was taught in the 1940s with how I begin treatment now in the 1990s. I argue that my present method is less rigid, less authoritarian and more realistic than the ‘classical’ procedure I was taught 50 years ago. I end the paper with some reflections on training analyses and their economic and emotional significance for both training analyst and trainee analysand.