The interplay of slow dynamics and thermodynamic features of dense liquids is studied by examinining how the glass transition changes depending on the presence or absence of Lennard-Jones-like attractions. Quite different thermodynamic behavior leaves the dynamics unchanged, with important consequences for high-pressure experiments on glassy liquids. Numerical results are obtained within mode-coupling theory (MCT), but the qualitative features are argued to hold more generally. A simple square-well model can be used to explain generic features found in experiment. The quest for identifying the physical mechanism behind the dynamical transformation of a liquid into an amorphous solid, the glass transition, has prompted many studies aiming to disentangle the dominant control variables involved. It is recognized that the slow dynamics connected with the glass transition is universal, but its connection to the underlying liquid structure is highly debated [1,2,3,4,5]. Some argue in terms of a density effect called free or excluded volume; others attribute the main physics to energetic interactions and thermally activated processes.Experiments changing both temperature T and pressure P close to the transition are emerging to resolve such issues, but have brought contradictory results. Some find that temperature dominates glassy dynamics by far [6,7,8,9]; some that it does not [10,11,12]. Others find both variables to exert equal influence [13,14,15,16,17,18,19,20,21,22], some with temperature [23,24,25,26,27], some with density ̺ being more relevant [28,29,30,31]. The difficulty of obtaining data over wide pressure ranges might be impeding: few studies, pioneered only in the 1990's [28,32], go beyond 1 GPa.Here we propose that to resolve the apparent contradictions, one needs to separate non-universal thermodynamic aspects, namely the equation of state (EOS) of the system, from universal dynamical features, viz. the slow relaxation. Specifically, we show how the presence or absence of attractive interactions affects the glass transition, and how this emerges in different pairs of variables linked by the EOS: (̺, T ) (preferred by theory) vs. (P, T ) (more amenable to experiments), yielding a transition that appears 'temperature-driven' in the latter.The Lennard-Jones (LJ) potential serves as a realistic interaction model: V LJ (r) = 4ǫ[(r/σ) −12 −(r/σ) −6 ], with dimensionless parameters ̺ * = ̺σ 3 , T * = 1/(βǫ), P * = P σ 3 /ǫ; β = 1/(k B T ) with Boltzmann's constant k B . To study relative effects of entropy and energy, we compare the LJ glass transition with that of its purely repulsive (LJR) counterpart, V LJR (r) = V LJ (r) + ǫ ≥ 0 for r ≤ 2 1/6 σ, V LJR (r) = 0 else. For the transition lines, modecoupling theory (MCT) [33,34] provides a reasonable qualitative description. Its transition temperature T c is systematically above the experimental (calorimetric) one, T g , but nevertheless serves as a good indicator for the change from high-T liquid-like dynamics to low-T glasslike one [35,36]. In MCT, T c is the...