The appropriate choice of shear strength of liquefied sands is an impor- tant component in seismic slope stability evaluation. Several factors affect the undrained steady-state strength (S us ) of sands. The steady-state strengths of 24 sandy soils were analyzed. It is shown that fines content, relative density, and friction angle play important roles affecting S us . Fines content was found to be the major factor affecting S us . This was verified experimentally for one sand. When the S us data for sands were grouped into (a) relatively clean sands (12 percent fines), (b) silty sands (12 to 50 percent fines), and (c) silts or sandy silts (50 percent fines), at the same relative density, relatively clean sands showed the highest S us . Silts showed the lowest S us . Silty sands showed intermedi- ate strengths. Lower-bound S us -relative density relationships were estab- lished for relatively clean sands and silty sands. The appropriate choice of shear strength of liquefied sands is an important component in seismic slope stability evaluation. It is well recognized that several factors may significantly affect the in situ postliquefaction shear strength of sands. At present, it is mostly determined by three methods: (a) steady-state strength (S us ) testing in the laboratory, followed by making appropriate corrections to the field void ratio condition taking relevant factors into account (1), (b) back-calculated residual strength (S r ) correlation with equivalent clean sand standard penetration test blow count [( N 1 ) 60-CS ] on the basis of past case studies (2,3), and (c) the normalized strength ratio approach (4–6). The S us approach for seismic stability analysis is based on the assumption that seismic flow deformation is affected primarily by one mechanism, based on the concept of steady-state defor- mation (1), except for possible differences due to the shearing mode. If the back-calculated S r data are also affected only by the latter mechanism, then practically, S r and S us should be the same, except for possible differences due to stress path effects. In a real field problem several mechanisms ( 2,7–9) may influence the operative strength. The back-calculated S r (2) may collectively reflect the effects of different mechanisms, if present, in each case history. At present the relationship between S r and S us or their relationship with postliquefaction strength is not clear. In general, past studies ( 2) indicate that the back-calculated residual strengths are smaller than typical laboratory S us values obtained for different sands at compara- ble relative densities. It is suspected that different mechanisms that might have been operative under given site conditions could have affected the residual strength, accordingly resulting in a different S r compared with the S us (2). There is concern among practitioners that the use of S us determined from laboratory tests may be unsafe (10). Having observed such differences, in general, there is consensus among researchers that within the limitations of current understand- ing of the factors affecting the postliquefaction strengths of sands, TRANSPORTATION RESEARCH RECORD 1547 61 the back-calculated S r correlations allow a judicious choice of postliquefaction strength for practical applications. Yet the choice of appropriate shear strength from S r versus (N 1 ) 60-CS correlations is a challenging task for a practicing engineer. Better understanding of the factors affecting S us and its possible relation to back-calculated residual strength would be valuable in selecting the appropriate strength for many practical problems. This paper presents an engineering reassessment of S us data for many sands. The objective is to identify the primary factors affect- ing S us and to delineate their roles. The focus is on friction angle, fines content, and relative density (D r ). Fines content (percentage passing through a No. 200 seive) is shown to be a major factor affecting S us . When the S us data for sands are grouped into (a) relatively clean sands (12 percent fines), (b) silty sands (12 to 50 percent fines), and (c) silts or sandy silts (50 percent fines), at the same D r , relatively clean sands show the highest S us . Silts show the lowest S us . Silty sands show intermediate strengths. On the basis of these data lower-bound S us values are established for each of these categories. It is also noted that the soils involved in many of the case histo- ries (2) consisted of significant amounts of fines. It appears that typ- ically lower back-calculated S r s (compared with the S us s of rela- tively clean sands at similar relative densities) observed in the case histories may be attributed to the effects of fines as one of the pri- mary causes. LABORATORY S us DATA The soil characteristics for a total of 24 different sandy soils involved in the study and their index properties [maximum void ratio (e max ), minimum void ratio (e min ), steady-state friction angle ( s ), particle sizes at 10 and 50 percent passing ( d 10 and d 50 , respec- tively), uniformity coefficient (C u ; C u = d 60 /d 10 ), angularity, fines content] and mechanical characteristics are summarized in Table 1. The S us data were obtained from e versus  3f ' space available in the literature and were converted to S us by Equation 1 when it was rel- evant: (1) and where q = ( 1 - 3 )  2, q = principal stress difference at the steady state,  3f '= effective minor principal stress at the steady state, = state friction angle,  1 = major principal stress, and  3 = minor principal stress. S q us f = ′ - = σ φ φ φ φ 3 1 sin cos sin cos Steady-State Strength, Relative Density, and Fines Content Relationship for Sands S. THEVANAYAGAM, K. RAVISHANKAR, AND S. MOHAN Department of Civil and Environmental Engineering, Polytechnic Univer- sity, 6 Metrotech Center, Brooklyn, N.Y. 11201.