Endotoxins are cell wall components of Gram-negative bacteria. A release of endotoxins into the human blood stream results in an inflammation reaction that can lead to life-threatening conditions like sepsis. Therefore, control for endotoxin contamination of intravenously administered drugs is crucial. Drugs are usually tested for putative endotoxin contamination with Limulus-based tests. However, validity of the compendial test procedures is questioned in the case of low endotoxin recovery (LER). To assure validity, regulatory authorities request hold-time studies of endotoxin in addition to pharmacopoeial requirements. Within these studies, endotoxin is added (spiked) to an undiluted product. The spiked product is held for a certain period of time and subsequently diluted for endotoxin determination. Due to the known heterogeneity of endotoxin the question has been raised as to which source represents the most adequate endotoxin spike. In the present study, endotoxin hold-time studies were analyzed by using different sources of endotoxin. Highly purified endotoxin, crude endotoxin extracts (Naturally Occurring Endotoxin) from different bacterial species and varied growth conditions as well as endogenous endotoxin contaminations were investigated. The results clearly demonstrate that endotoxin masking—an effect of LER—is dependent on the endotoxin source used. Various parameters such as bacterial strain and growth conditions lead to different masking susceptibilities. Due to these effects it is impossible to predict the susceptibility of bacterial endotoxin contamination to LER. In order to determine whether a sample is prone to LER, an endotoxin spike that is susceptible to LER is required.