Activity cliffs occur in drug discovery programmes when a small structural change leads to a very large change in biological activity. We have used a matched molecular pair approach for common functional groups to establish examples of activity cliffs. We then measure how positive and negative activity cliffs are over-represented relative to a normal distribution. The ‘Magic Methyl Effect’ highlighted examples of where methylation led to significant activity increases, yet our data suggests that negative activity cliffs are more common than positive ones for methyl, and that significant increases are less common than most other functional groups. We show that both scaffold and substitution position effects play an important role in the occurrence of activity cliffs.