In Arabidopsis thaliana, a consistent multiloop clock model has been widely adopted in many recent publications. This tentative model consists of three interactive feedback loops, namely the core CCA1/LHY-TOC1/X loop, the morning CCA1/LHY-PRR9/PRR7 loop and the evening Y-TOC1 loop, in which the undefined Y gene might be GI. The model in its current form provides us with a basis on which to address a number of fundamental issues for a better understanding of the molecular mechanism by which the central oscillator generates circadian rhythms. We have been conducting a series of genetic studies through the establishment of a set of combinatorial mutants. We have already characterized a prr9 prr7 double loss-of-function mutant that has lost the morning loop, and a cca1 lhy toc1 triple mutant that lacks the core loop. Extension of this line of study required characterization of a gi toc1 double loss-of-function mutant, which is expected to have no evening loop, and a prr9 prr7 toc1 triple mutant, lacking both the morning and evening loops. Genetic analysis of both these lines is reported here. From the results, we have clarified the genetic linkages between GI and TOC1 and those between PRR9/PRR7 and TOC1 with reference to the circadian clock-associated phenotypes, including: (i) length of hypocotyls during early photomor-phogenesis; (ii) photoperiodic control of flowering time; and (iii) expression profiles of CCA1 and LHY under free-running conditions. These results indicate that GI is not sufficient to fulfill the Y role, but plays more complicated clock-associated roles and, interestingly, that no epistatic interaction between PRR9/PRR7 and TOC1 was observed. Furthermore, these clock-defective mutants could still generate robust, free-running rhythms at the level of transcription. Therefore, we speculate that an as yet undefined oscillator (or loop) continues to generate rhythms within the plants lacking GI/TOC1 or PRR9/PRR7/TOC1.