||Currently many evaluation models on the photobiological effects (PBE) of light sources do not consider the influence of age and luminance on the pupil diameter, which affects the light radiation intensity on the human retina. In this study, the pupil diameter is taken into consideration when evaluating the PBE of several light sources. Moreover, the correction factor M is proposed. The blue light hazard (BLH) efficacy and the circadian rhythm (CR) effects of the daylight at dusk, together with three indoor light sources with a correlated color temperature (CCT) of about 3000 K were evaluated by using a corrected evaluation model. The results show that an incandescent lamp is more photobiologically friendly for humans, despite being inefficient. Based on high wall-plug efficiency (WPE) GaN-based yellow (565 nm, 24.3%@20 A/cm2) and green (522 nm, 41.3%@20 A/cm2) LEDs on silicon substrate, incandescent-like spectrum and phosphor-free color-mixed white LEDs (CM-LEDs) with a general color rendering index (CRI) of 94, a CCT of 2866 K, and an efficiency of 131 lm/W were manufactured by mixing blue, cyan, green, yellow and red LEDs. The PBE evaluation results of such CM-LEDs are superior to those of an incandescent lamp. Moreover, blue light free and candlelight-toned LEDs with an efficiency of 120.3 lm/W, a general CRI of 84, a special CRI R9 of 93.3, and a CCT of 1810 K were fabricated by mixing yellow and red LEDs (R&Y-mixed LEDs). The R&Y-mixed LEDs show no blue light weighted quantities and have a weaker influence on the CR shift. They are photobiologically friendly for humans and suitable for nocturnal indoor and outdoor lighting environments.