Small-angle neutron scattering (SANS) experiments were performed on Zr 41.2Ti 13.8Cu 12.5Ni 10Be 22.5 (Vit1) and Zr 46.8Ti 8.2Cu 7.5Ni 10Be 27.5 (Vit4), as well as on three other compositions (Vit1A, B and C), where the (Zr, Ti) and (Cu, Be) composition (following the line in composition space connecting Vit1 and Vit4) was varied. The SANS data of the Vit1 type alloys, annealed at 623 K, show interference peaks, giving evidence for spatially correlated arrangements of inhomogeneities. The Q values of these peaks, Q max, decrease along the connecting line from Vit1 to Vit4. SANS data of Zr 42.6Ti 12.4Cu 11.25Ni 10Be 23.75 (Vit1A) further show that Q max decreases with increasing annealing temperature. From an in situ SANS experiment, performed on Vit1A for 15 h at 621 K, we derive a quantitative model for the crystallization pathway of Vit1 type alloys. We propose that, in the early stages, these types of alloys decompose without crystallization. In the later stages, decomposition and crystallization occur simultaneously. The nucleation and growth of nanocrystals in the decomposed glassy matrix results in relaxation toward a new chemical equilibrium state, in which the size of the nanocrystals is determined by the length scale of the initial decomposition of the amorphous matrix. In this case, conventional power law crystal growth theories completely fail. Instead the experimental data can be with a relaxation function over the whole time interval. A diffusion constant and Kohlrausch exponent, obtained from fitting the data, are in agreement with values obtained from independent diffusion and viscosity measurements.