An SRAM array and method of making is disclosed. Each SRAM cell comprises two pull-up (PU), two pass-gate (PG), and two pull-down (PD) FinFETs. The PU transistors are adjacent to each other and include one active fin having a first fin width. Each PG transistor shares at least one active fin with a PD transistor. The at least one active fin shared by a PG and a PD transistor has a second fin width smaller than the first fin width. The method includes patterning a plurality of fins including active fins and dummy fins and patterning and removing at least a portion of the dummy fins. No dummy fin is disposed between PU FinFETs in a memory cell. One dummy fin is disposed between a PU FinFET and the at least one active fin shared by a PG and a PD transistor. At least one dummy fin is disposed between adjacent memory cells.