Strength measurements are becoming increasingly important for electroceramics. Bending of specimens small enough to be cut out of small electroceramic components may be one possibility. Therefore the miniaturisation of the 4-point bend-test for ceramic specimens is now being attempted. In this paper the errors in determining the flexural strength arising from the test principle itself, plus the geometry and measuring inaccuracies are calculated and expressed as a function of the outer span length. Contact pressure and a tolerable total measuring inaccuracy determines the dimensions of miniature specimens and fixtures. The possibilities of appropriate specimen preparation are also investigated.Ceramic materials show a volume (i.e. a specimen size) dependence of strength which is described by Weibull's statistical theory. The applicability of the miniature bend-fixtures is demonstrated by measuring this volume effect. NOMENCLATURE h = specimen's breadth F = load at failure h = specimens height rn = Weibull exponent p =contact pressure r = roller radius E l , E , = Young's moduli of two different materials 1, L = inner and outer span length M , = maximum bending moment PF = probability of failureVo, V, = volume, normalising volume, effective volume W, , , = maximum section modulus of bending specimen x, y, z = rectangular coordinates E, E,, cran = total error, error caused by one single source, random error v = Poisson's ratio (r = stress pk, pr = coefficient of kinetic and rolling friction (ro, ny = characteristic strength, yield strength (rb, ox,, = maximum flexural stress from simple beam theory and actual stress