Target selection is a critical step in the majority of modern drug discovery programs. The viability of a drug target depends on two components: biological relevance and chemical tractability. The concept of druggability was introduced to describe the second component, and it is defined as the ability of a target to bind a drug-like molecule with a therapeutically useful level of affinity. To investigate the rules that govern druggability, we developed an algorithm to isolate and characterize the binding pockets of protein targets. Using this algorithm, we performed a comparative analysis between the relevant pockets of 60 targets of approved drugs and a diverse set of 440 ligand-binding pockets. As a result, we defined a preferred property space for druggable pockets based on five key properties (volume, depth, enclosure, percentage of charged residues and hydrophobicity), and we represented it with a set of simple rules. These rules may be applicable in the future to evaluate the chemical tractability of prospective targets.