This paper presents the results of an experimental programme to investigate the performance of a new composite shear connector that can be used to build deconstructable composite floors. A total of 21 push-out tests were conducted to quantify the effects of different design parameters, including different nominal temperature levels (20 °C, 100 °C, 300 °C, 450 °C, 600 °C), with the steel decking perpendicular and parallel to the steel section, two different embedded shear stud lengths and two different grades of concrete (C40/50 and C20/25). The tests were mainly carried out under steady-state condition where the specimen temperatures were increased to the desired target and the load was then applied until failure of the specimen. A few transient tests were also carried out, in which a load equal to the load carrying capacity of the steady-state test at 600 °C was applied, followed by increasing the specimen temperature until failure. Three modes were observed: (i) concrete crushing with stud fracture, (ii) stud fracture and (iii) concrete pull-out. Concrete pull-out occurred only on one specimen with concrete grade C20/25. It is found that the main failure mode at elevated temperatures not exceeding 400 °C was initiated by concrete crushing, concomitantly producing ductile stud damages. At elevated temperatures above 400 °C, the failure mode changed to shear stud fracture. The Eurocode 4 calculation equations were shown to produce shear connector resistances to be in reasonable agreement with the steady state results for the higher-grade concrete, even though the previous equations were developed for welded shear studs. The transient state tests failed at lower temperatures than the target temperature of 600 °C, indicating it may not be safe to use steady state tests for evaluation of shear connector resistance at high temperatures.