Pedicle screw loosening has been implicated in recurrent back pain after lumbar spinal fusion, but the degree of loosening has not been systematically quantified in patients. Instrumentation removal is an option for patients with successful arthrodesis, but remains controversial. Here, we quantified pedicle screw loosening by measuring screw insertion and/or removal torque at high statistical power (β = 0.98) in N = 108 patients who experienced pain recurrence despite successful fusion after posterior instrumented lumbar fusion with anterior lumbar interbody fusion (L2-S1). Between implantation and removal, pedicle screw torque was reduced by 58%, indicating significant loosening over time. Loosening was greater in screws with evoked EMG threshold under 11 mA, indicative of screw misplacement. A theoretical stress analysis revealed increased local stresses at the screw interface in pedicles with decreased difference in pedicle thickness and screw diameter. Loosening was greatest in vertebrae at the extremities of the fused segments, but was significantly lower in segments with one level of fusion than in those with two or more. Clinical significance: These data indicate that pedicle screws can loosen significantly in patients with recurrent back pain and warrant further research into methods to reduce the incidence of screw loosening and to understand the risks and potential benefits of instrumentation removal.