The present paper by Pai et al. [1] is a well-motivated attempt to improve dosage regimens for continuous intravenous meropenem administration in morbidly obese patients with stable or unstable renal function who have severe Gram-negative bacterial infections. The desire to employ continuous intravenous infusion of time-dependent antibiotics to maximize their therapeutic effect is laudable. The authors appear to have used a nomogram based on a formula they previously described for meropenem clearance [2]. They then correlated meropenem clearance with creatinine clearance (CLCR). In the present paper by Pai et al. [1], they obtained what are described as steady-state serum samples of meropenem in the obese patients they studied. They made a one-compartment pharmacokinetic model of meropenem. They then calculated individual Bayesian estimates of the meropenem central volume of distribution (Vc) and clearance (CL) in each obese patient. From this, they developed a dosage nomogram for various categories of renal function, designed for their obese patients. This is a step in the right direction, since assays for meropenem are not commonly available. However, in their previous paper [2], their target steady-state concentration was 8–12 lg/mL. Figure 2 in their previous paper [2] describes the serum concentrations observed in validating the previous nomogram. They ranged from about 2 to about 19 lg/mL. Counting the points in that plot, about 47 % were below the target range of 8–12 lg/mL, about 29 % were within it and about 26 % were above it. This seems to be the evidence they used to validate their earlier nomogram [2]. In that paper, they stated that they gave a loading dose, which they described as essential. However, in the present paper by Pai et al. [1], I could not find mention of a loading dose. In Figs. 1 and 2 of the present paper by Pai et al. [1], the assumed steady-state serum concentrations range from about 2 to about 100 lg/mL. The authors report 41.1 % of serum concentrations as being over 16 lg/mL (the apparent upper limit of their target range), 76.1 % as being at least 8 lg/mL (so only 35 % were within the target range of 8–16 lg/mL), 97.4 % being at least 4 lg/mL (so 21 % within 4–8 lg/mL) and 99.9 % being at least 2 lg/mL (or 2.5 % within 2–4 lg/mL). So only 35 % were within 8–16 lg/mL, while 65 % were outside the target range. The very large range of serum concentrations found actually suggests that either (1) there was large interpatient variability; (2) the serum concentrations might not have been at steady state; or (3) the previous nomogram was not very well suited to the present obese patients. The authors seem not to have seriously considered an upper limit of serum concentrations but mostly to have set a lower therapeutic threshold as the target concentration. If this is so, then one apparently might give almost any large enough dosage regimen without consideration of toxicity. While the incidence of toxicity appears to be small, seizures have been reported, and one might consider setting some upper constraint on the serum concentrations to be achieved. The authors state that ‘‘the base structural model provided an excellent fit to the data’’, as in their Fig. 2. & Roger Jelliffe jelliffe@usc.edu; http://www.lapk.org