To study improved oil recovery (IOR) via laboratory experiments at the pore scale, we performed waterflooding experiments in a glass 2.5D micromodel (dual depth: 12 and 27 μm) with crude oil (CRO) and brines of variable compositions at temperatures ranging from 22 °C (room temperature) to 90 °C. The time-dependent residual oil saturation (ROS) for various flooding and aging protocols was extracted from optical microscopy images of the entire pore space in the micromodel. Additionally, we used high-resolution images to examine the microscopic distributions of oil and brine at the subpore level. Variation of the fluid aging history (before the first flooding with high-salinity water, HSW) revealed that sequential aging with formation water and CRO led to significantly higher ROS values than aging with CRO only. Video analysis of the pore space showed that most of the oil was trapped via a complete bypassing of the deep pores. On increasing the waterflooding temperature, both the ROS and the fraction of bypassed pores became smaller. An increase in dewetting of tiny oil drops and films from the pore walls supports the notion of a ROS decrease via a wettability alteration. Subsequent flooding with low-salinity water (LSW) did not lead to recovery of additional oil, regardless of aging condition or temperature. Our results show the significance of fluid aging and temperature to design a successful microfluidic IOR strategy.