Lg, a regional seismic wave comprised of multiple shear wave reverberations trapped in the crustal waveguide, is important for magnitude estimation and source discrimination for monitoring nuclear testing treaties. In stable continental regions, Lg propagates with a group velocity of about 3.5 km/s and can often be observed at distances up to 4000 km. To better understand the absence of high-frequency Lg arrivals for paths traversing the northern boundary of the Tibetan Plateau, we investigate spatial variations of broadband (0.15--5.0 Hz) energy in the Lg group velocity window (3.6--3.0 km/s) using regional waveforms recorded at the Chinese Digital Seismic Network station WMQ. Vertical component seismograms are analyzed for 90 events with magnitudes of 4.4 ≤ mb ≤ 6.4 that occurred between 1987 and 1999 in the Tibetan Plateau and around its margins. The Lg amplitude spectra for events located near the northern margin of the plateau have apparent corner frequencies of 1--2 Hz, nearly identical to those for comparable size events at similar distances outside the plateau. High-frequency (>1 Hz) Lg energy recorded at WMQ decreases rapidly as a function of source distance into the plateau. A path length of 300--400 km within the northern Plateau suffices to eliminate 2--5 Hz Lg energy. For events in southern Tibet with paths crossing the central portion of the Tibetan Plateau, almost total Lg extinction occurs, even for energy in the low-frequency band of 0.2--1 Hz. Corresponding apparent corner frequencies of the Lg amplitude spectra range between 0.2 and 0.4 Hz. The corner frequency shift is found to vary systematically with path length across the plateau. Linear regressions demonstrate that the shift in apparent corner frequency of Lg amplitude spectra is negatively correlated with features of the Tibetan Plateau, such as mean elevation along the paths or travel distance within the plateau above specified elevation thresholds. The systematic variations in the amplitude and frequency content of energy in the Lg window as a function of path length within the plateau indicate that strong crustal attenuation plays an important role in Lg extinction for paths traversing central and northern Tibet, superimposed on any structural blockage effects associated with abrupt thinning of the crust near the northern boundary of the plateau. Spectral ratios of many event pairs along great circle paths give estimates of frequency-dependent Lg attenuation for paths crossing western, central, and eastern sectors of Tibet. The region of strong Sn attenuation in northern central Tibet also has strong Lg attenuation with QLg for 1 Hz on the order of Q0 = 80--90, while in southern central Tibet, Q0 increases to about 316, and in eastern and western Tibet, Q0 is on the order of 120--200 for paths traversing the entire plateau. The strong Lg attenuation in northern central Tibet is responsible for the so-called Lg blockage and may be associated with partial melting in the crustal low-velocity layer in northern Tibet. Our QLg values for Tibet are significantly lower than most earlier estimates, primarily as a result of not excluding blocked observations along with allowing for lateral variations within Tibet.