Abstract Background Temocillin plasma protein binding (PPB) in healthy individuals is reported to be ∼85% but had not been studied in patients. Objectives To obtain normative data on temocillin PPB in patients in relation to infection and impact of co-medications widely used in ICU. Methods Plasma was obtained from healthy individuals (Group #1), non-ICU patients with UTI (Group #2), ICU patients with suspected/confirmed ventriculitis (Group #3) or with sepsis/septic shock (Group #4). Total and unbound temocillin concentrations were measured in spiked samples from temocillin-naive donors (in vitro) or in plasma from temocillin-treated subjects (in vivo). The impact of diluting plasma, using pharmaceutical albumin, or adding drugs potentially competing for PPB was tested in spiked samples. Data were analysed using a modified Hill–Langmuir equation taking ligand depletion into account. Results Temocillin PPB was saturable in all groups, both in vitro and in vivo. Maximal binding capacity (Bmax) was 1.2–2-fold lower in patients. At 20 and 200 mg/L (total concentrations), the unbound fraction reached 12%–29%, 23%–42% and 32%–52% in Groups #2, #3, #4. The unbound fraction was inversely correlated with albumin and C-reactive protein concentrations. Binding to albumin was 2–3-fold lower than in plasma and non-saturable. Drugs with high PPB but active at lower molar concentrations than temocillin caused minimal displacement, while fluconazole (low PPB but similar plasma concentrations to temocillin) increased up to 2-fold its unbound fraction. Conclusions Temocillin PPB is saturable, 2–4-fold lowered in infected patients in relation to disease severity (ICU admission, hypoalbuminaemia, inflammation) and only partially reproducible with albumin. Competition with other drugs must be considered for therapeutic concentrations to be meaningful.