The effects testing of pharmaceuticals consists of a tiered investigation of ecotoxicological endpoints. However, effects testing has to be performed only when the predicted environmental concentrations (PECs) of pharmaceuticals are above certain action limits. To study the appropriateness of these action limits, a literature search was performed for pharmaceuticals with predicted no-effect concentrations (PNECs) close to or below the action limits. Some human pharmaceuticals showed effects at concentrations ≤100 ng/L, mostly in nonstandard fish or invertebrate tests. In addition, antibiotics and parasiticides sometimes had effects at concentrations <10 mg/kg soil. To help in identifying pharmaceuticals that should undergo effects testing although their PECs are below the action limits, "however clauses" are postulated for pharmaceuticals that are potentially persistent, bioaccumulative, carcinogenic, mutagenic, or reproductively toxic. Effects testing should also be performed for pharmaceuticals that 1) affect target structures that are conserved across species, 2) have a high potency or a small therapeutic margin, 3) are from a new therapeutic class, and 4) are structurally similar to compounds with known effects. Furthermore, suggestions for improving the effects testing of pharmaceuticals are made. These include inter alia chronic effects testing as a general approach, the use of invertebrate tests including sexual reproduction, the application of endpoints reflecting the mode of action of the drug or known side effects, and the simulation of more realistic exposure conditions in terrestrial laboratory tests.