In practice, the load-curve and distributed generation (DG) penetration level determines the power imbalance level that a network can experience if islanding occurs. Therefore, with the prior knowledge of load-curve and DG penetration level, the operating point or setpoint of Rate-of-change-of-frequency (ROCOF) relays can be adjusted so as to make them suitable for a real network. To this end, this paper firstly investigates the subcritical power imbalance region of ROCOF relays through analytical formulation; then, simulation study is conducted under different load-types and at the existence of deficit/excess of active and/or reactive power imbalance scenarios in order to establish the maximum boundary limit of ROCOF under all scenarios. Secondly, employing reliability (assessed by Detection Rate and False Alarm Rate) and response time as key performance indicators, ROCOF's performance is evaluated using numerous islanding and nonislanding events simulated from a test network of Australia. Lastly, ROCOF's performance is assessed using the ROC (Receiver Operating Characteristics) curve along with relay setpoint; the assessment considers the number of islanding events associated with the time-wise percentage of power imbalance level (deficit/excess) calculated from the different DG penetration level and daily load-curve. All these test results demonstrate a clear guideline to select ROCOF's setpoint.