OBJECTIVE To evaluate the stepwise approach in differentiating between benign and malignant salivary gland tumors using time-intensity curves (TICs) and apparent diffusion coefficients (ADCs). METHODS TICs and ADCs were analyzed on the tumor-by-tumor (overall) and pixel-by-pixel (TIC and ADC maps) bases in patients with benign (n = 52) or malignant (n = 18) salivary gland tumor. TICs were categorized into Types 1 (<20% increment ratio), 2 (>or=20% increment ratio and >120 sec peak time), 3 (>or=20% increment ratio, <or=120 sec peak time, and <30% washout ratio), or 4 (>or=20% increment ratio, <or=120 sec peak time, and >or=30% washout ratio). ADCs were classified as extremely low (<0.6 x 10(-3) mm(2)/sec), low (<1.2), intermediate (<1.8), or high (>or=1.8). RESULTS Malignant tumors had small (<30%) areas with Type 1 TIC with one of the following magnetic resonance imaging (MRI) characteristics: Type 3 overall TIC patterns, Type 4 overall TIC patterns and extremely low (<0.60 x 10(-3) mm(2)/sec) overall ADCs, or Type 2 overall TIC patterns and large (>40%) areas with low or extremely low ADCs. CONCLUSIONS We propose a stepwise approach by using multiparametric MRI techniques as an effective tool for differentiating between benign and malignant salivary gland tumors.