The question of why different experiments obtain different shapes of Landau subbands of two-dimensional electron gases (2DEG's) is investigated. In particular, we consider disordered graphene within the tight-binding formulation in a high magnetic field and in the presence of impurity potentials of a finite interaction range. It is found that the shape of the total density of states (DOS) of Landau subbands is well described by a Gaussian function, while the shape of the local density of states (LDOS) can be fitted to either a simple Lorentzian function or a sum of several Lorentzian functions. This finding explains well why the experiments that measure the DOS of a Landau subband appear to be Gaussian, and those experiments that are sensitive to the LDOS can only be fitted well by Lorentzian functions. Thus, our results provide a natural explanation to this long-standing puzzle involving 2DEG's in the quantum Hall regime. Graphene has attracted a lot of attention due to its remarkable properties and its potential applications in nanoelectronics. 1 Its electronic properties are mainly due to its two-dimensional nature and the fact that the low-energy excitations are governed by the massless Dirac equation, 2 confirmed by the interesting, uneven Landau level (LL) distribution that varies with both the square root of the magnetic field B and the LL index n: E n = sgn(n) 2ehv 2 F |n|B, where v F , e, andh are the Fermi velocity, electron charge, and the Planck constant, respectively. This emblematic LL spectrum and its characteristic zero-energy state (n = 0) are directly related to the anomalous quantum Hall effect (QHE) in graphene. 3 Many efforts have been made to determine the electronic properties of the disorder-broadened Landau subbands (LS's). However, the shape of both the total density of states (DOS) and the local DOS (LDOS) of an LS of graphene is a controversial issue in the literature. Historically, it has generally been believed that the shape of the DOS of LS's in a conventional two-dimensional electron gas (2DEG) is Gaussian-like, 4,5 but experiments seem to present a different picture. Initially, this consensus was supported by careful measurements of specific heat, 6 magnetization, 7 and magnetocapacitance 8 on GaAs-Ga x Al 1−x As heterostructures. However, a recent scanning-tunneling-microscopy-scanningtunneling-spectroscopy (STM-STS) measurement on conventional 2DEG's of the n-InSb(110) ] surface, 9 as well as the cyclotron resonance (CR) absorption line shape on conventional 2DEG's, 10,11 observed a Lorentzian shape of an LS. The situation for disordered graphene is similar. All CR 13-15 and STM-STS 16-19 measurements suggested clearly a Lorentzian shape for an LS. However, recent magnetocapacitance experiments produced conflicting results: In one experiment, 20 an LS could be fitted well by the Lorentzian functions, while another experiment 21 observed a Gaussian shape for an n = 0 LS. On the theoretical side, the DOS of an LS of graphene is initially found to be Gaussian-like. [22][23][24] C...