Doctor of Philosophy / Department of Physics / Chii-Dong Lin / Imaging of molecules has always occupied an essential role in physical, chemical and biological sciences. X-ray and electron diffraction methods routinely achieve sub-angstrom spatial resolutions but are limited to probing dynamical timescales longer than a picosecond. With the advent of femtosecond intense lasers, a new imaging paradigm emerges in last decade based on laser-induced electron diffraction (LIED). It has been placed on a firm foundation by the quantitative rescattering theory, which established that large-angle e-ion elastic differential cross sections (DCS) can be retrieved from the LIED spectrum. We further demonstrate that atomic potentials can be accurately retrieved from those extracted DCSs at energies from a few to several tens of electron volts. Extending to molecules, we show mid-infrared (mid-IR) lasers are crucial to generate high-energy electron wavepackets (> 100 eV) to resolve the atomic positions in a molecule. These laser-driven 100 eV electrons can incur core-penetrating collisions where the momentum transfer is comparable to those attained in conventional keV electron diffraction. Thus a simple independent atom model (IAM), which has been widely used in conventional electron diffractions, may apply for LIED. We theoretically examine and validate the applicability of IAM for electron energies above 100 eV using e-molecule large-angle collision data obtained in conventional experiments, demonstrating its resolving powers for bond lengths about 0.05 angstrom. The Validity of IAM is also checked by an experimental LIED investigation of rare gas atoms in the mid-IR regime. We show that the electron’s high energy promotes core-penetrating collisions at large scattering angles, where the e-atom interaction is dominated by the strong short range atomic-like potential. Finally, we analyze the measured LIED spectrum of N[subscript]2 and O[subscript]2 at three mid-IR wavelengths (1.7, 2.0, and 2.3 μm). As expected, the retrieved bond lengths of N[subscript]2 at three wavelengths are about same as the equilibrium N[subscript]2 bond length. For O[subscript]2, the data is also consistent with a bond length contraction of 0.1 angstrom within 4-6 fs after tunnel ionization. This investigation establishes a foundation for this novel imaging method for spatiotemporal imaging of gas-phase molecules at the atomic scale.