Bacterial cells behave as individuals despite being genetically identical and subject to the same environment. Although the underlying mechanisms of cellular individuality are not well understood, temporal variation of gene expression and protein levels in individual cells could act as sources for phenotypic heterogeneity of populations. Bacterial persistence in fluctuating environments relies on such diversity, allowing some individuals, by chance, to survive a challenge that would kill an unadapted and phenotypically homogeneous population. Bacterial persistence is a non-genetic, metastable phenomenon that is readily lost upon subculture of surviving cells in the absence of the environmental stress. A medically relevant example of persistence is the fractional killing of bacteria by antibiotics, whereby a subpopulation of cells survives indefinitely despite prolonged exposure to a bactericidal drug. It is hypothesized in the case of tuberculosis (TB) that drug therapy may require administration of multiple drugs for 6-9 months in order to eliminate a subpopulation of "persister" cells. The focus of this thesis is to understand how persisters tolerate antibiotics that kill their genetically identical siblings. To address this question, automated time-lapse fluorescence microcopy was applied in conjunction with microfluidic and microelectromechanical systems (MEMS) to study bacterial responses to antibiotics in real time at the single-cell level. A genetic approach was employed to identify mutants of Mycobacterium smegmatis with altered rates of persistence against the first-line anti-TB drug isoniazid (INH). A microfluidic platform was designed for high-throughput screening of transposon (Tn) mutant libraries. A pilot screen of 1,100 Tn mutants revealed 11 mutants that showed increased (persister-up, PU) or decreased (persister-down, PD) survival when exposed to isoniazid. These mutants were further characterized in comparison to wild-type cells using the single-cell microfluidics platform. This single-cell analysis revealed distinct modes of persistence within each category, which were otherwise masked at the population level. The results indicate that persistence behavior was the cumulative result of the dynamics of cell division and cell lysis. Mutants with decreased persistence exhibited behavior at single-cell level that challenges the accepted mechanism of isoniazid-mediated killing. Therefore, these results yield new insights into antibiotic persistence and isoniazid's mode of action. Persisters cannot be easily studied in non-fractionated populations of cells where they represent a small percentage of the total. As a complementary approach, a method was developed based on dielectrophoretic (DEP) purification of naturally occurred low frequency persister cells from an antibiotic-treated bacterial culture. Purification of persister cells would allow for downstream analysis using conventional bulk/population level transcriptomic and proteomic approaches. The crossover frequencies for live and dead cells were determined to enable selective trapping or release from the device. This approach enabled a proof-of-concept enrichment of live bacteria in the mixture of live and dead cells after INH treatment, based on their dielectric properties. Quantitative confirmation of the enriched/isolated cell populations was attempted using flow cytometry. The high-throughput microfluidic screening and single-cell approaches developed in this thesis could, in principal, be applied to the genetic analysis of a wide variety of dynamic cellular processes that can be imaged by time- lapse fluorescence microscopy, such as gene expression, cell division, chromosome replication and segregation, and protein localization. The biochemical analysis of live cells isolated by dielectrophoresis-based separation would provide useful information about the underlying persistence mechanism. Dielectrophoresis could also be a valuable analytical tool for mutant characterization based on their intrinsic dielectric properties, either at single-cell or population levels. In conclusion, this thesis presents several new microfabricated tools to identify and understand bacterial persistence. The thesis illustrates how integration of microfluidic tools with automated time-lapse microscopy allows hidden characteristics that are present but invisible in the macro- world of batch cultures to be examined in a quantitative and high-throughput manner in the micro-world of single-cell analysis.