Almost all reported intramolecular [2 + 2] reactions of ene-keteniminium ions gave normal [2 + 2] products with a fused bicycle framework, but not cross [2 + 2] products with a bicyclo[3.1.1]heptane skeleton, a highly pursued bioisostere in pharmaceutical chemistry. How to rationalize this and design new cross [2 + 2] reactions? Theoretical studies using density functional theory, high-level ab initio single-point energy calculations, and molecular dynamics showed that this [2 + 2] reaction has all three patterns of regiochemical control: the reaction is controlled either kinetically, thermodynamically, or dynamically. A carbocation model of forming endo and exo carbocations has been proposed to rationalize the reaction outcomes, revealing that the tethers (between alkenes and keteniminium ions), substituents (on the alkenes), and alkene configurations in ene-keteniminium ions play critical roles. These understandings were further used to predict that introducing a substituent in the terminal position of alkene with a trans configuration in eneketeniminium ions can realize the cross [2 + 2] reaction, which is dynamically controlled for alkyl substituents or kinetically controlled for aryl substituents. These and more other predictions were realized experimentally, and many cross [2 + 2] products with a bicyclo[3.1.1]heptane skeleton can be achieved. Both molecular dynamics and new experiments have also been applied to correct a key but misassigned [2 + 2] product reported in the literature, further supporting the insightful mechanisms reported here.