Until now, most research concerning the atmosphere’s structure in the Inn Valley (Austria), based on vertical profiles of temperature and humidity, is made with selected case studies or numerical model simulations. Nowadays, ground based passive microwave radiometers can provide this profile information in a high temporal sampling resolution. In this study, data from such an instrument (HATPRO) is used to investigate the valley atmosphere structure evolution in terms of stratification. Associated mixed layer (ML) heights (z vM L ) and stable boundary layer (SBL) heights (z vSBL ) within or above the valley, which correspond in parts to the atmospheric boundary layer height (ABL-height) in a flat horizontally homogeneous (HHF) case, were determined. Case studies for the two main types of boundary layers, the convective boundary layer (CBL), including a ML, and the SBL, show that the temperature profiles can present some of the typical CBL-features, while typical SBL-features in the corresponding profiles are mostly missing. The humidity profiles provide the right order of magnitude and vertical gradient, but almost no fine scale structures. Only some of the investigated methods to detect z vM L /z vSBL yield reasonable results compared to the theoretical knowledge from HHF-cases. Based on this quality assessment, seasonally averaged diurnal cycles of virtual potential temperature (θ v ) profiles and the whole data set of variously detected z vM L /z vSBL were further investigated. In the mean spring-, summer- and autumn-profiles a ML evolves during day mainly by turbulent heat flux from the valley floor. The atmosphere up to the mean crest height is additionally heated by recirculating slope winds from the valley side walls. The mean winter profiles are mostly difficult to examine in terms of CBL evolution, as the θ v -stratification stays stable over the whole “mean-”day. For spring and summer, the mean profiles of the evolved convective valley atmosphere show a stable layer above the actual mixed layer, followed by a near-mixed layer around crest height, followed by the stable free atmosphere. The most promising detection methods for z vM L are the “parcel method” and the “parameterization method”. Both show clearly a diurnal cycle and find the highest ML-heights during spring and summer in the afternoon. Extreme values of the parcel method even reach higher altitudes than the mean crest height. The detection of z vSBL seems to be very difficult and further research is needed.