Health studies based on noise-induced effects are usually conducted utilizing the outdoor noise exposure levels rather than the indoor exposures. In addition to this, studies make no distinction based on the sound insulation behaviour of façade structures and only a few studies have reported spectral characteristics. Given that low-frequency spectral components of night-time noise exposure can disrupt activities such as sleep, it is important to estimate outdoor-to-indoor attenuation factors with respect to low-frequency ranges. This study evaluates the outdoor-to-indoor spectral characteristics based on night-time outdoor and indoor noise measurements at 49 locations in Greater London, in the UK. Since the measurements were conducted using an unsupervised approach for almost three consecutive days, the data were processed to eliminate the presence of dominant secondary sound sources and to minimize the contribution of indoor background noise. For the latter verification, two background noise criteria were defined: 6dBA and 10dBA. By analysing the data in relation to two different time-interval steps (1-hour and 8-hours), it was found that the sound insulation is underestimated for the less stringent criterion in certain 1/3-octave frequency bands. This behaviour was minimised using the more stringent criterion. Considering 1/1-octave bands, this underestimation of sound insulation was significantly eliminated. Focusing on the low-frequency components of nighttime noise exposure (50Hz-250Hz), multi and mixed-effects linear regression models were used to estimate outdoor-to-indoor noise levels based on predictors associated with outdoor noise levels, outdoor environment, indoor environment, and building façade characteristics, reaching RMSE levels of 3.8dBA-2.8dBA and 1.7dBA-0.7dBA, respectively. Finally, attenuation factors for estimating the level of internal low-frequency components of noise exposure based on external ones were extracted from these models. The factors achieved attenuation levels around 20dBA.