Mercury intrusion porosimetry (MIP) is a widely used method for studying porous materials, in particular, cement-based materials. The usual interpretation of such measurements is based on certain assumptions. One of these is that each pore is connected to the sample surface directly or through larger pores. Pores not meeting this assumption are called ink-bottle pores. The effect that sample size has on the MIP characteristics of concrete samples, like the ink-bottle effect and hysteresis, was studied by measuring additional extrusion and intrusion cycles. In order to characterize the extrusion and ink-bottle behavior, the amount of entrapped mercury chi(p) was estimated. Superimposition of extrusion and second intrusion curves is achieved if the contact angle theta is adjusted from theta(i), the intrusion contact angle, to theta(e), the extrusion contact angle. The threshold radius is often assumed to be a dominant pore radius, yet in this study the entrapped mercury content shows no evidence for the presence of a dominant pore radius. Even if characteristic properties of cement-based materials can be estimated with MIP, comparison of results is rendered difficult by the significant effects of sample preparation techniques and sample size and the ink-bottle effect due to randomly present air bubbles.