Applied classical measurement theory was used to study the measurement characteristics of Forms A and B of the Finding Embedded Figures Test (FEFT) when the test is administered in a "no-guessing" or "supply" format. Data provided by 69 students at a private university in the southern United States were used. Both forms of the FEFT were administered in counterbalanced order to different subjects; 36 subjects completed Form A, and then Form B. Specifically, the study was designed to compare: (1) the alpha coefficient reliabilities of data from the two administration formats; (2) test and item difficulty data across administration methods; and (3) corrected item-total correlation of discrimination coefficients across test administrations. Results are contrasted with those of a previous study involving a multiple-choice "selection" format administration of the FEFT to 302 subjects. The alpha correlation associated with scores of the 69 subjects on Form A was 0.66 and that for Form B was 0.83. In terms of test difficulty, Form A scores ranged from 16 through 32, and Form B scores ranged from 13 through 34. Form A items were more likely to behave differently across administration formats. Both the selection- and supply-format administrations of each form had positive discrimination coefficients. The two FEFT forms provide data with reasonable reliability and psychometric integrity. The FEFT may assist researchers who want to use a selection format to measure field independence. (TJH)