The withdrawal timing of direct oral anticoagulants (DOACs) among patients in need of elective invasive surgery is based on DOAC pharmacokinetics in order to perform the procedure out of the DOAC peak plasma concentration. We aimed to investigate the prevalence and predictors of plasma levels of DOACs out of trough range in patients with atrial fibrillation (AF) in need of elective cardiac procedure. We evaluated all consecutive AF patients on DOAC therapy in need of elective cardiac procedure, admitted to our division from January 2022 to March 2022. All patients underwent DOAC plasma dosing the morning of procedure day. They were categorized as in range, above range, and below range, according to the DOAC reference range at the downstream point. The timing of discontinuation of DOAC therapy was considered as appropriate or not, according to the current recommendations. The clinical predictors of out-of-range DOAC plasma levels have been evaluated. We included 90 consecutive AF patients (56.6% male, mean age 72.95 ± 10.12years); 74 patients (82.22%) showed DOAC concentration out of the expected reference range. In half of them (n, 37), the DOAC plasma concentration was below the trough reference range. Of the study population, 17.7% received inappropriate DOAC dosages (10% overdosing, 7% underdosing), and 35.5% had incorrect timing of DOAC withdrawal (26% prolonged, 9.5% shortened). At multivariable analysis, inappropriate longer DOAC withdrawal period (OR 10.13; P ≤ 0.0001) and increased creatinine clearance (OR 1.01; P = 0.0095) were the independent predictors of plasma DOAC levels below the therapeutic trough range. In contrast, diabetes mellitus (OR 4.57; P = 0.001) was the only independent predictor of DOAC plasma level above the therapeutic trough range. Increased creatinine clearance and inappropriate longer drug withdrawal period are the only independent predictors of DOAC plasma levels below the reference range; in contrast, diabetes is significantly correlated with DOAC plasma levels above the reference.