Procrastination is a familiar concept with rich hidden implications for learning science. Although the effect of procrastination on learner success has been reported extensively in the research literature, attempts to quantify it remain elusive. In this paper we investigate the relationship between procrastination and academic performance. Our goal is to quantify the relationship as set of probabilities: what is the precise cost (as risk) of delaying learning tasks to achievement or academic outcomes? Our dataset consists of 2.6 million observations spanning 105,000 students, 60,000 unique assignments in 2,500 course sections. Our investigation shows that habitually delaying learning tasks leads to adverse outcomes to the point of catastrophe. This in itself is not a surprising result. But we confirm the adage that good students have good habits and poor students have poor habits, thereby lending support to the under-appreciated view that meta-cognitive traits play a vital role in learner success.