The aim of the present study was to compare the technique currently used for carrying out the microtensile bond strength (muTBS) test with an alternative method. Six human molars were sectioned transversally to obtain two specimens from each tooth: a 0.75-mm-thick slab of mid-coronal dentin, which presented an occlusal and a cervical surface (DS specimen); and the remaining section of the tooth with an exposed dentin surface (SS specimen). A self-etch adhesive system and four layers (2 mm each) of resin composite were applied to the sectioned surfaces by means of a custom-made special device, in order to obtain a cylindrical build-up. Each specimen was sectioned lengthwise to obtain multiple sticks. The end sections of sticks obtained from DS specimens were resin composite, whereas the middle section was dentin. The sticks obtained from SS specimens were half resin composite and half dentin. The specimens were fractured under tension. Data were subjected to a multilevel statistical model. The maximum load at break of SS specimens was higher than that of DS specimens. This modified methodology offers two main advantages over the traditional procedure: easier manipulation of the sticks; and the possibility of testing deep dentin.