Safety margin, a key aspect of any non-clinical toxicity studies, is calculated by dividing the systemic exposure (AUC) at NOAEL (No Adverse Effect Level) in toxicity studies by the clinical exposure. The validity of using total plasma concentration (C_p) to calculate AUC is often discussed, as it is the unbound plasma concentration (C_up) that elicits the pharmacological and toxicological effects. Data regarding plasma protein binding across species was collected for 114 MSD small molecule compounds which had been discontinued from development either due to non-clinical toxicity or due to clinical Adverse Effects. A >3-fold difference in unbound fraction in plasma (fu_p) was selected as a meaningful difference in plasma protein binding between non-clinical species and humans. In rats, dogs and non-human primates, approximately 3–5% of the compounds had a >3-fold difference in plasma protein binding than humans. Following assessment of toxicity profile of these compounds, it was concluded that calculation of safety margins after incorporating fu_p would have still led to the discontinuation of these compounds. Therefore, although fu_p can still be used for calculation of safety margin on a case by case basis, the routine use of fu_p for calculation of safety margins is not warranted.