Research has indicated that students can be taught self-regulated learning (SRL) in scaffolding programs focusing on a fixed continuous practice (e.g., metacognitive question prompts). However, the fading role of scaffolding to prepare autonomous learning is often an overlooked component. A unique approach for fading is suggested that offers a graduated reduction model of scaffolding prompts according to the SRL phases involved in the solution, which allows assimilation of processes to prepare learners for autonomous activity. This quasi-experimental study of fourth-graders (n = 134) examines the effectiveness of metacognitive self-question prompts in a Fixed (continuous) versus Faded (graduated reduction) scaffolds model during planning, monitoring and reflection phases, on the facilitation of students' SRL (metacognition, calibration of confidence judgment, motivation), and sense making of mathematical problem solving at the end of the program (short-term effect) and 3 months later (long-term/lasting effect). Findings indicated that the Faded Group performed best in the metacognition knowledge aspect, motivation in the performance goal approach increased and, in the avoidance, goal decreased. No differences were found between the groups on the regulation aspect and calibration of confidence judgment in the solution success. Additionally, the Faded Group outperformed the Fixed Group on sense making of problem solving. These findings were manifested particularly in the long-term effect. The study supports theoretical claims relating the role of fading scaffolds to increase students' autonomous SRL (metacognition, motivation) and improvements in sense making, particularly on the long-term retention effect.