Astronomical (or Milanković) forcing of the Earth system is key to understanding rhythmic climate change on time scales ≳104 y. Paleoceanographic and paleoclimatological applications concerned with past astronomical forcing rely on astronomical calculations (solutions), which represent the backbone of cyclostratigraphy and astrochronology. Here we present state‐of‐the‐art astronomical solutions over the past 3.5 Gyr. Our goal is to provide tuning targets and templates for interpreting deep‐time cyclostratigraphic records and designing external forcing functions in climate models. Our approach yields internally consistent orbital and precession‐tilt solutions, including fundamental solar system frequencies, orbital eccentricity and inclination, lunar distance, luni‐solar precession rate, Earth's obliquity, and climatic precession. Contrary to expectations, we find that the long eccentricity cycle (LEC) (previously assumed stable and labeled “metronome,” recent period ∼405 kyr), can become unstable on long time scales. Our results reveal episodes during which the LEC is very weak or absent and Earth's orbital eccentricity and climate‐forcing spectrum are unrecognizable compared to the recent past. For the ratio of eccentricity‐to‐inclination amplitude modulation (recent individual periods of ~2.4 and ~1.2 Myr, frequently observable in paleorecords) we find a wide distribution around the recent 2:1 ratio, that is, the system is not restricted to a 2:1 or 1:1 resonance state. Our computations show that Earth's obliquity was lower and its amplitude (variation around the mean) significantly reduced in the past. We therefore predict weaker climate forcing at obliquity frequencies in deep time and a trend toward reduced obliquity power with age in stratigraphic records. For deep‐time stratigraphic and modeling applications, the orbital parameters of our 3.5‐Gyr integrations are made available at 400‐year resolution.