Tankyrases-1 and -2 (TNKS-1 and TNKS-2) have three cellular roles which make them important targets in cancer. Using NAD+ as a substrate, they poly(ADP-ribosyl)ate TRF1 (regulating lengths of telomeres), NuMA (facilitating mitosis) and axin (in wnt / beta-catenin signalling). Using molecular modelling and the structure of the weak inhibitor 5-aminoisoquinolin-1-one, 3-aryl-5-substituted-isoquinolin-1-ones were designed as inhibitors to explore the Structure-Activity Relationships (SARs) for binding and to define the shape of a hydrophobic cavity in the active site. 5-Amino-3-arylisoquinolinones were synthesised by Suzuki-Miyaura coupling of arylboronic acids to 3-bromo-1-methoxy-5-nitroisoquinoline, reduction and O-demethylation. 3-Aryl-5-methylisoquinolin-1-ones, 3-aryl-5-fluoroisoquinolin-1-ones and 3-aryl-5-methoxyisoquinolin-1-ones were accessed by deprotonation of 3-substituted-N,N,2-trimethyl¬benzamides and quench with an appropriate benzonitrile. SAR around the isoquinolinone core showed that aryl was required at the 3-position, optimally with a para-substituent. Small meta-substituents were tolerated but groups in the ortho-positions reduced or abolished activity. This was not due to lack of coplanarity of the rings, as shown by the potency of 4,5-dimethyl-3-phenylisoquinolin-1-one. Methyl and methoxy were optimal at the 5-position. SAR was rationalised by modelling and by crystal structures of examples with TNKS-2. The 3-aryl unit was located in a large hydrophobic cavity and the para-substituents projected into a tunnel leading to the exterior. Potency against TNKS-1 paralleled potency against TNKS-2. Most inhibitors were highly selective for TNKSs over PARP-1 and PARP-2. A range of highly potent and selective inhibitors is now available for cellular studies.