Iododeoxyuridine (IUdR) uptake and retention was imaged by positron emission tomography (PET) at 0-48 min and 24 h after administration of 28.0-64.4 MBq (0.76-1.74 mCi) of [124I]IUdR in 20 patients with brain tumors, including meningiomas and gliomas. The PET images were directly compared with gadolinium contrast-enhanced or T2-weighted magnetic resonance images. Estimates for IUdR-DNA incorporation in tumor tissue (Ki) required pharmacokinetic modeling and fitting of the 0-48 min dynamically acquired data to correct the 24-h image data for residual, nonincorporated radioactivity that did not clear from the tissue during the 24-h period after IUdR injection. Standard uptake values (SUVs) and tumor:brain activity ratios (Tm:Br) were also calculated from the 24-h image data. The Ki, SUV, and Tm/Br values were related to tumor type and grade, tumor labeling index, and survival after the PET scan. The plasma half-life of [124I]IUdR was short (2-3 min), and the arterial plasma input function was similar between patients (48 +/- 12 SUV*min). Plasma clearance of the major radiolabeled metabolite ([124I]iodide) varied somewhat between patients and was markedly prolonged in one patient with renal insufficiency. It was apparent from our analysis that a sizable fraction (15-93%) of residual nonincorporated radioactivity (largely [124I]iodide) remained in the tumors after the 24-h washout period, and this fraction varied between the different tumor groups. Because the SUV and Tm:Br ratio values reflect both IUdR-DNA incorporated and exchangeable nonincorporated radioactivity, any residual nonincorporated radioactivity will amplify their values and distort their significance and interpretation. This was particularly apparent in the meningioma and glioblastoma multiforme groups of tumors. Mean tumor Ki values ranged between 0.5 +/- 0.9 (meningiomas) and 3.9 +/- 2.3 microl/min/g (peak value for glioblastoma multiforme, GBM). Comparable SUV and Tm:Br values at 24 h ranged from 0.13 +/- 0.03 to 0.29 +/- 0.19 and from 2.0 +/- 0.6 to 6.1 +/- 1.5 for meningiomas and peak GBMs, respectively. Thus, the range of values was much greater for Ki (approximately 8-fold) compared with that for SUV (approximately 2.2-fold) and Tm:Br (approximately 3-fold). The expected relationships between Ki, SUV, and Tm:Br and other measures of tumor proliferation (tumor type and grade, labeling index, and patient survival) were observed. However, greater image specificity and significance of the SUV and Tm:Br values would be obtained by achieving greater washout and clearance of the exchangeable fraction of residual (background) radioactivity in the tumors, i.e., by increased hydration and urinary clearance and possibly by imaging later than 24 h after [124I]IUdR administration.